import tensorflow as tf
# from tensorflow.python.client import device_lib

gpus = tf.config.list_physical_devices('GPU')
if gpus:
  try:
    # Currently, memory growth needs to be the same across GPUs
    for gpu in gpus:
      tf.config.experimental.set_memory_growth(gpu, True)
    logical_gpus = tf.config.list_logical_devices('GPU')
    print(len(gpus), "Physical GPUs,", len(logical_gpus), "Logical GPUs")
  except RuntimeError as e:
    # Memory growth must be set before GPUs have been initialized
    print(e)

2 Physical GPUs, 2 Logical GPUs

import os
import sys
import gc
#import cv2
import numpy as np
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Input, Conv3D, MaxPooling3D, Flatten, Dense, LSTM, TimeDistributed, GlobalAveragePooling2D, MaxPooling2D
from tensorflow.keras.applications.resnet50 import ResNet50
from tensorflow.keras.utils import to_categorical
#from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
from tensorflow.keras.utils import plot_model
import keras.utils as image

from tensorflow.keras.callbacks import EarlyStopping
from keras.optimizers import Adam
from keras.models import Sequential
from keras.callbacks import ReduceLROnPlateau
from sklearn.metrics import confusion_matrix
from tensorflow.keras.callbacks import EarlyStopping
from keras.callbacks import ModelCheckpoint
from sklearn.metrics import classification_report

tf.test.is_built_with_cuda()

True

gc.collect()

0

# Define constants
CATEGORIES = ['dining', 'drawing', 'mopping_floor', 'reading_book', 'running_on_treadmill','sleeping','watching_tv']
NUM_CLASSES = len(CATEGORIES)
INPUT_SHAPE = (40, 112, 112, 3)  # (frames, height, width, channels)
BATCH_SIZE = 4
EPOCHS = 30
random_state = 42
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'

def read_dataset_40f(date_dir, extractor):
    X_train = np.load(date_dir + extractor + '/Split/X_train.npy')
    print('Loaded X_train')
    y_train = np.load(date_dir + extractor + '/Split/y_train.npy')
    print('Loaded y_train')
    X_test = np.load(date_dir + extractor + '/Split/X_test.npy')
    print('Loaded X_test')
    y_test = np.load(date_dir + extractor + '/Split/y_test.npy')
    print('Loaded y_test')
    
    print("Shape of X_train: ", X_train.shape)
    print("Shape of y_train: ", y_train.shape)
    print("Shape of X_test: ", X_test.shape)
    print("Shape of y_test: ", y_test.shape)
    
    gc.collect()
    
    return X_train, y_train, X_test, y_test

def read_evaldataset_40f(extractor):
    X_eval = np.load('/home/ycaun/Desktop/jeheng/Processed dataset/Validate_data_processed/' + extractor + '/' + extractor + '_X_eval.npy')
    y_eval = np.load('/home/ycaun/Desktop/jeheng/Processed dataset/Validate_data_processed/' + extractor + '/' + extractor + '_y_eval.npy')

    print(X_eval.shape)
    print(y_eval.shape)
    
    gc.collect()
    
    return X_eval, y_eval

#### Model setup:
Extractor = 'resnet_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

X_train, y_train, X_test, y_test = read_dataset_40f('resnet_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

history = model.fit(X_train, y_train, batch_size=BATCH_SIZE, epochs=EPOCHS, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-04-18 11:18:35.527245: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.
2023-04-18 11:18:37.602430: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.

Epoch 1/30

2023-04-18 11:18:53.765125: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:428] Loaded cuDNN version 8401
2023-04-18 11:18:56.920197: I tensorflow/compiler/xla/service/service.cc:173] XLA service 0x7ff2881584c0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-04-18 11:18:56.920292: I tensorflow/compiler/xla/service/service.cc:181]   StreamExecutor device (0): GeForce GTX 1080, Compute Capability 6.1
2023-04-18 11:18:58.055101: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:268] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-04-18 11:19:02.633020: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 11:19:02.633103: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 11:19:02.633204: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 11:19:02.633324: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.

   1/2193 [..............................] - ETA: 14:26:01 - loss: 2.1391 - accuracy: 0.0000e+00

2023-04-18 11:19:03.030656: I tensorflow/compiler/jit/xla_compilation_cache.cc:477] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.
2023-04-18 11:19:03.189016: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.

2191/2193 [============================>.] - ETA: 0s - loss: 1.8363 - accuracy: 0.2427

2023-04-18 11:19:11.438239: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.
2023-04-18 11:19:11.917350: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.

Epoch 1: val_loss improved from inf to 1.78711, saving model to lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

X_eval, y_eval = read_evaldataset_40f('resnet_gavg')

(678, 40, 2048)
(678, 7)

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

170/170 [==============================] - 0s 3ms/step

(678, 7)

confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1))

array([[30,  1,  4, 26, 19,  5, 13],
       [ 3, 49, 12,  9,  0, 16,  3],
       [ 7,  6, 26, 12,  9, 12, 28],
       [11,  4, 10, 28, 12, 14, 19],
       [14,  3, 12, 17, 24,  6, 20],
       [ 7,  7, 10, 22,  2, 38, 10],
       [ 8,  2, 10,  9,  6,  6, 57]])

print(classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

              precision    recall  f1-score   support

           0       0.38      0.31      0.34        98
           1       0.68      0.53      0.60        92
           2       0.31      0.26      0.28       100
           3       0.23      0.29      0.25        98
           4       0.33      0.25      0.29        96
           5       0.39      0.40      0.39        96
           6       0.38      0.58      0.46        98

    accuracy                           0.37       678
   macro avg       0.39      0.37      0.37       678
weighted avg       0.38      0.37      0.37       678

#### Model setup:
Extractor = 'vgg16_avg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_avg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30
2193/2193 [==============================] - ETA: 0s - loss: 1.1955 - accuracy: 0.5702
Epoch 1: val_loss improved from inf to 1.05269, saving model to lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 23s 6ms/step - loss: 1.1955 - accuracy: 0.5702 - val_loss: 1.0527 - val_accuracy: 0.6312 - lr: 0.0010
Epoch 2/30
2181/2193 [============================>.] - ETA: 0s - loss: 0.8919 - accuracy: 0.6907
Epoch 2: val_loss improved from 1.05269 to 0.95942, saving model to lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 12s 6ms/step - loss: 0.8910 - accuracy: 0.6912 - val_loss: 0.9594 - val_accuracy: 0.6522 - lr: 0.0010
Epoch 3/30
2187/2193 [============================>.] - ETA: 0s - loss: 0.7790 - accuracy: 0.7337
Epoch 3: val_loss improved from 0.95942 to 0.93662, saving model to lstm_vgg16_avg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_avg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 0s 2ms/step

(678, 7)

confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1))

array([[59,  1,  5, 18,  4,  6,  5],
       [ 1, 64, 10,  6,  5,  6,  0],
       [10,  4, 73,  0,  9,  1,  3],
       [ 9, 11,  8, 54,  5,  7,  4],
       [ 6,  4,  8, 13, 59,  1,  5],
       [ 0,  3,  7, 21,  2, 60,  3],
       [ 3,  2,  7, 12,  7,  3, 64]])

print(classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

              precision    recall  f1-score   support

           0       0.67      0.60      0.63        98
           1       0.72      0.70      0.71        92
           2       0.62      0.73      0.67       100
           3       0.44      0.55      0.49        98
           4       0.65      0.61      0.63        96
           5       0.71      0.62      0.67        96
           6       0.76      0.65      0.70        98

    accuracy                           0.64       678
   macro avg       0.65      0.64      0.64       678
weighted avg       0.65      0.64      0.64       678

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8768, 40, 512)
Shape of y_train:  (8768, 7)
Shape of X_test:  (2192, 40, 512)
Shape of y_test:  (2192, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 128)               328192    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 329,095
Trainable params: 329,095
Non-trainable params: 0
_________________________________________________________________

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30
2189/2192 [============================>.] - ETA: 0s - loss: 1.2433 - accuracy: 0.5417
Epoch 1: val_loss improved from inf to 1.01082, saving model to lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 17s 5ms/step - loss: 1.2431 - accuracy: 0.5420 - val_loss: 1.0108 - val_accuracy: 0.6432 - lr: 0.0010
Epoch 2/30
2183/2192 [============================>.] - ETA: 0s - loss: 0.9558 - accuracy: 0.6637
Epoch 2: val_loss improved from 1.01082 to 0.91480, saving model to lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 10s 4ms/step - loss: 0.9560 - accuracy: 0.6635 - val_loss: 0.9148 - val_accuracy: 0.6743 - lr: 0.0010
Epoch 3/30
2189/2192 [============================>.] - ETA: 0s - loss: 0.8256 - accuracy: 0.7088
Epoch 3: val_loss improved from 0.91480 to 0.86492, saving model to lstm_vgg16_gavg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 512)
(678, 7)
170/170 [==============================] - 1s 1ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[69  1  2 10  2  8  6]
 [ 2 74  1  6  6  1  2]
 [ 7  3 71  2 12  2  3]
 [13  9  6 55  4  7  4]
 [ 6  4  8  9 60  3  6]
 [ 3  4  6 15  6 60  2]
 [ 7  0  7 13  6  2 63]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.64      0.70      0.67        98
           1       0.78      0.80      0.79        92
           2       0.70      0.71      0.71       100
           3       0.50      0.56      0.53        98
           4       0.62      0.62      0.62        96
           5       0.72      0.62      0.67        96
           6       0.73      0.64      0.68        98

    accuracy                           0.67       678
   macro avg       0.67      0.67      0.67       678
weighted avg       0.67      0.67      0.67       678

#### Model setup:
Extractor = 'vgg16_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir,'vgg16_gavg',)

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 512)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 512)
Shape of y_test:  (2480, 8)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 128)               328192    
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 329,224
Trainable params: 329,224
Non-trainable params: 0
_________________________________________________________________

2023-09-12 01:45:44.349669: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:45:44.350596: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:45:44.351179: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-12 01:45:50.849808: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 813137920 exceeds 10% of free system memory.
2023-09-12 01:45:51.266118: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 813137920 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 01:45:51.661283: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:45:51.662258: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:45:51.662924: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:45:52.024557: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:45:52.025395: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:45:52.026194: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:45:52.404250: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-12 01:45:52.449801: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f53300e96d0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-12 01:45:52.449821: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 01:45:52.449825: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 01:45:52.454004: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-12 01:45:52.564558: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2469/2482 [============================>.] - ETA: 0s - loss: 1.3327 - accuracy: 0.5277

2023-09-12 01:46:02.096123: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:02.097120: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:02.097735: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.16490, saving model to lstm_vgg16_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-12 01:46:03.308116: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:03.309054: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:03.309710: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:03.425726: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:03.426447: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:03.427082: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:03.534783: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:03.535554: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:03.536286: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:03.553687: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.558767: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.662960: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:03.663651: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:03.664294: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:03.681029: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,512]
	 [[{{node input_1}}]]
2023-09-12 01:46:03.781427: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:03.782354: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:03.783007: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:03.798528: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,512]
	 [[{{node input_1}}]]
2023-09-12 01:46:03.806083: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.811865: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,512]
	 [[{{node input_1}}]]
2023-09-12 01:46:03.818077: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.823509: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,512]
	 [[{{node input_1}}]]
2023-09-12 01:46:03.829102: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.833720: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.843097: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.848418: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.882664: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,512]
	 [[{{node input_1}}]]
2023-09-12 01:46:03.893658: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:03.899095: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:04.000666: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.001582: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.002214: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:04.230328: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.231027: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.231732: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:04.261701: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,512]
	 [[{{node inputs_0}}]]
2023-09-12 01:46:04.266422: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,512]
	 [[{{node inputs_0}}]]
2023-09-12 01:46:04.271620: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:04.276424: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,512]
	 [[{{node inputs}}]]
2023-09-12 01:46:04.377679: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.378362: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.378997: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:04.482955: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.483831: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.484459: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:04.593630: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.594318: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.594932: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:46:04.701817: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:46:04.702703: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:46:04.703336: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 01:46:04.764423: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,512]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_vgg16_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

2482/2482 [==============================] - 14s 5ms/step - loss: 1.3313 - accuracy: 0.5283 - val_loss: 1.1649 - val_accuracy: 0.5944 - lr: 0.0010
Epoch 2/30
2467/2482 [============================>.] - ETA: 0s - loss: 1.0091 - accuracy: 0.6481
Epoch 2: val_loss improved from 1.16490 to 1.09508, saving model to lstm_vgg16_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 512)
(778, 8)
 77/195 [==========>...................] - ETA: 0s

2023-09-12 01:51:40.233716: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:51:40.234395: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:51:40.235097: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 0s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[65  0  2 13  3  8  6  1]
 [ 1 70  3  7  4  4  1  2]
 [ 6  5 65  3  8  3  3  7]
 [11  8  2 49  6 10  8  4]
 [ 3  3 12  9 54  1  6  8]
 [ 5  0  6 21  7 50  6  1]
 [ 4  0  8  7  7  6 61  5]
 [ 6  3  6  7 11  3  8 56]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.64      0.66      0.65        98
           1       0.79      0.76      0.77        92
           2       0.62      0.65      0.64       100
           3       0.42      0.50      0.46        98
           4       0.54      0.56      0.55        96
           5       0.59      0.52      0.55        96
           6       0.62      0.62      0.62        98
           7       0.67      0.56      0.61       100

    accuracy                           0.60       778
   macro avg       0.61      0.60      0.61       778
weighted avg       0.61      0.60      0.61       778

#### Model setup:
Extractor = 'mobV2_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1280)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1280)
Shape of y_test:  (2191, 7)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 722,311
Trainable params: 722,311
Non-trainable params: 0
_________________________________________________________________

2023-09-09 00:15:16.385668: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:15:16.386565: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:15:16.387158: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-08 04:41:46.159100: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 1795891200 exceeds 10% of free system memory.
2023-09-08 04:41:47.155056: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 1795891200 exceeds 10% of free system memory.

Epoch 1/30

2023-09-08 04:41:47.902888: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:47.903971: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:47.904656: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:48.269543: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:48.270487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:48.271147: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:48.641718: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-08 04:41:48.711972: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f10ac02eca0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-08 04:41:48.711998: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 04:41:48.712002: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 04:41:48.739899: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.

  17/2193 [..............................] - ETA: 7s - loss: 2.0245 - accuracy: 0.2206

2023-09-08 04:41:48.995333: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2192/2193 [============================>.] - ETA: 0s - loss: 0.8491 - accuracy: 0.7100

2023-09-08 04:41:57.467389: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 448716800 exceeds 10% of free system memory.
2023-09-08 04:41:57.733988: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 448716800 exceeds 10% of free system memory.
2023-09-08 04:41:58.009340: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:58.010091: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:58.010804: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.70976, saving model to lstm_mobV2_gavg_128_unit_4_Adam_ReduceLROnPlateau

2023-09-08 04:41:59.353508: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:59.354475: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:59.355089: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:59.472306: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:59.472993: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:59.473628: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:59.583243: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:59.584129: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:59.584768: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:59.602756: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.607987: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.714320: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:59.715009: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:59.715645: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:59.733362: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 04:41:59.836008: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:41:59.836925: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:41:59.837577: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:41:59.853666: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 04:41:59.861501: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.867787: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 04:41:59.874155: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.879593: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 04:41:59.885272: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.890004: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.900296: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.905828: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.941094: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 04:41:59.952774: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:41:59.958640: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:42:00.177204: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.178097: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.178719: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:42:00.289479: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.290179: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.290802: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:42:00.322683: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-08 04:42:00.327696: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-08 04:42:00.332948: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:42:00.337703: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 04:42:00.441181: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.441999: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.442609: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:42:00.550764: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.551462: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.552095: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:42:00.666287: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.667126: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.667759: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:42:00.775766: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:42:00.776428: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:42:00.777036: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-08 04:42:00.840232: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1280)
(678, 7)
 20/170 [==>...........................] - ETA: 0s

2023-09-08 04:48:53.134068: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 04:48:53.134964: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 04:48:53.135593: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 04:48:53.228969: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

170/170 [==============================] - 1s 3ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[77  1  3  9  5  2  1]
 [ 1 78  2  7  1  3  0]
 [ 5  2 77  7  4  2  3]
 [ 9 12  2 66  0  6  3]
 [ 3  4  7  7 70  2  3]
 [ 2  6  6 18  3 58  3]
 [ 1  2  7 13  5  3 67]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.79      0.79      0.79        98
           1       0.74      0.85      0.79        92
           2       0.74      0.77      0.75       100
           3       0.52      0.67      0.59        98
           4       0.80      0.73      0.76        96
           5       0.76      0.60      0.67        96
           6       0.84      0.68      0.75        98

    accuracy                           0.73       678
   macro avg       0.74      0.73      0.73       678
weighted avg       0.74      0.73      0.73       678

#### Model setup:
Extractor = 'mobV2_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1280)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1280)
Shape of y_test:  (2480, 8)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 722,440
Trainable params: 722,440
Non-trainable params: 0
_________________________________________________________________

2023-09-12 01:11:13.278070: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:13.279010: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:13.279633: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-12 01:11:20.703524: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.
2023-09-12 01:11:21.712393: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 01:11:22.488562: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:22.489530: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:22.490173: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:22.854714: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:22.855590: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:22.856251: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:23.227933: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-12 01:11:23.301436: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7fee703a7320 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-12 01:11:23.301465: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 01:11:23.301472: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 01:11:23.306523: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-12 01:11:23.414529: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2476/2482 [============================>.] - ETA: 0s - loss: 0.9359 - accuracy: 0.6900

2023-09-12 01:11:33.454882: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:33.455815: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:33.456426: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.79080, saving model to lstm_mobV2_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-12 01:11:34.718954: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:34.719889: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:34.720535: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:34.831943: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:34.832654: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:34.833290: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:34.940193: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:34.941098: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:34.941746: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:34.960422: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:34.966412: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.077239: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.077954: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.078597: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:35.096059: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 01:11:35.197461: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.198521: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.199166: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:35.214822: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 01:11:35.222544: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.228387: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 01:11:35.234534: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.240115: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 01:11:35.245695: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.250390: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.259976: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.265403: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.300177: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 01:11:35.311268: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.316825: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.537321: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.537992: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.538623: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:35.646972: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.647823: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.648490: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:35.679378: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 01:11:35.684212: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 01:11:35.689284: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.693994: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 01:11:35.793117: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.793803: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.794426: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:35.899177: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:35.900151: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:35.900803: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:36.011763: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:36.012443: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:36.013048: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:11:36.121197: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:11:36.122129: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:11:36.122756: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 01:11:36.183970: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

2482/2482 [==============================] - 14s 5ms/step - loss: 0.9354 - accuracy: 0.6903 - val_loss: 0.7908 - val_accuracy: 0.7464 - lr: 0.0010
Epoch 2/30
2480/2482 [============================>.] - ETA: 0s - loss: 0.6389 - accuracy: 0.7888
Epoch 2: val_loss did not improve from 0.79080
2482/2482 [==============================] - 11s 4ms/step - loss: 0.6387 - accuracy: 0.7888 - val_loss: 0.8430 - val_accuracy: 0.7113 - lr: 0.0010
Epoch 3/30
2479/2482 [============================>.] - ETA: 0s - loss: 0.5589 - accuracy: 0.8122
Epoch 3: val_loss improved from 0.79080 to 0.76549, saving model to lstm_mobV2_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1280)
(778, 8)
 20/195 [==>...........................] - ETA: 0s

2023-09-12 01:17:54.451073: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 01:17:54.452074: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 01:17:54.452730: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 01:17:54.553653: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

195/195 [==============================] - 1s 3ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[71  2  0 14  4  0  1  6]
 [ 0 71  4  8  2  4  0  3]
 [ 0  4 66  5  6  2  3 14]
 [ 3  9  2 60  9  6  4  5]
 [ 0  2  5  7 70  1  4  7]
 [ 0  5  1 11  5 68  4  2]
 [ 3  1  3  8  6  6 66  5]
 [ 1  4  5  8  3  8  1 70]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.91      0.72      0.81        98
           1       0.72      0.77      0.75        92
           2       0.77      0.66      0.71       100
           3       0.50      0.61      0.55        98
           4       0.67      0.73      0.70        96
           5       0.72      0.71      0.71        96
           6       0.80      0.67      0.73        98
           7       0.62      0.70      0.66       100

    accuracy                           0.70       778
   macro avg       0.71      0.70      0.70       778
weighted avg       0.71      0.70      0.70       778

#### Model setup:
Extractor = 'mobV2_gavg_9class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (11078, 40, 1280)
Shape of y_train:  (11078, 9)
Shape of X_test:  (2768, 40, 1280)
Shape of y_test:  (2768, 9)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 9)                 1161      
                                                                 
=================================================================
Total params: 722,569
Trainable params: 722,569
Non-trainable params: 0
_________________________________________________________________

2023-09-09 04:08:17.097524: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:17.098264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:17.099038: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-09 04:08:21.600110: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2268774400 exceeds 10% of free system memory.
2023-09-09 04:08:22.773701: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2268774400 exceeds 10% of free system memory.

Epoch 1/30

2023-09-09 04:08:23.647884: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:23.648917: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:23.649614: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:24.038001: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:24.038986: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:24.039643: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:24.405007: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-09 04:08:24.455506: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f37b42e5d40 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-09 04:08:24.455537: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 04:08:24.455543: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 04:08:24.460326: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-09 04:08:24.568063: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2761/2770 [============================>.] - ETA: 0s - loss: 1.0043 - accuracy: 0.6776

2023-09-09 04:08:35.687685: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:35.688683: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:35.691268: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.90145, saving model to lstm_mobV2_gavg_9class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-09 04:08:37.304371: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:37.305671: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:37.306341: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:37.423859: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:37.424543: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:37.425165: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:37.534961: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:37.535816: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:37.536472: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:37.554394: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.559608: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.667411: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:37.668088: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:37.668712: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:37.686217: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-09 04:08:37.788712: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:37.789580: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:37.790198: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:37.806050: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-09 04:08:37.813849: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.819789: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-09 04:08:37.826053: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.831615: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-09 04:08:37.837324: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.842000: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.851533: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.856907: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.892138: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-09 04:08:37.903186: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:37.908708: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:38.128552: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.129313: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.130069: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:38.241162: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.241843: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.242451: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:38.274251: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-09 04:08:38.279555: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-09 04:08:38.284936: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:38.289800: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-09 04:08:38.391750: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.392617: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.393226: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:38.498465: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.499117: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.499711: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:38.611031: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.611814: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.612424: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:08:38.718823: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:08:38.719487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:08:38.720095: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-09 04:08:38.781684: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_9class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_9class_128_unit_4_Adam_ReduceLROnPlateau/assets

2770/2770 [==============================] - 16s 5ms/step - loss: 1.0033 - accuracy: 0.6780 - val_loss: 0.9015 - val_accuracy: 0.7124 - lr: 0.0010
Epoch 2/30
2768/2770 [============================>.] - ETA: 0s - loss: 0.6935 - accuracy: 0.7697
Epoch 2: val_loss improved from 0.90145 to 0.84930, saving model to lstm_mobV2_gavg_9class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(878, 40, 1280)
(878, 9)
 21/220 [=>............................] - ETA: 0s

2023-09-09 04:15:44.433355: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:15:44.434276: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:15:44.434899: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:15:44.528404: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

220/220 [==============================] - 1s 3ms/step

(878, 9)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[63  1  2 11  3  4  2  7  5]
 [ 1 62  2  7  0  7  0  8  5]
 [ 0  2 65  4  4  2  2 17  4]
 [ 1  8  2 58  2  6  4 14  3]
 [ 0  1  8  3 54  6  3 16  5]
 [ 0  1  3  6  4 72  1  6  3]
 [ 0  0  4  9  5  7 65  4  4]
 [ 1  3  5  4  4  2  3 78  0]
 [ 3  4  3 11  4 13  0  5 57]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.91      0.64      0.75        98
           1       0.76      0.67      0.71        92
           2       0.69      0.65      0.67       100
           3       0.51      0.59      0.55        98
           4       0.68      0.56      0.61        96
           5       0.61      0.75      0.67        96
           6       0.81      0.66      0.73        98
           7       0.50      0.78      0.61       100
           8       0.66      0.57      0.61       100

    accuracy                           0.65       878
   macro avg       0.68      0.65      0.66       878
weighted avg       0.68      0.65      0.66       878

#### Model setup:
Extractor = 'mobV2_gavg_10class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (11852, 40, 1280)
Shape of y_train:  (11852, 10)
Shape of X_test:  (2962, 40, 1280)
Shape of y_test:  (2962, 10)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 10)                1290      
                                                                 
=================================================================
Total params: 722,698
Trainable params: 722,698
Non-trainable params: 0
_________________________________________________________________

2023-09-08 23:59:06.138131: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:06.139067: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:06.139687: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-08 23:59:10.361275: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2427289600 exceeds 10% of free system memory.
2023-09-08 23:59:11.726567: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2427289600 exceeds 10% of free system memory.

Epoch 1/30

2023-09-08 23:59:12.706227: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:12.707692: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:12.708370: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:13.034974: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:13.035929: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:13.036640: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:13.339285: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-08 23:59:13.381233: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f3b0c7daf90 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-08 23:59:13.381258: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 23:59:13.381262: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 23:59:13.384486: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-08 23:59:13.489827: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2954/2963 [============================>.] - ETA: 0s - loss: 1.1637 - accuracy: 0.6261

2023-09-08 23:59:24.383925: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:24.384923: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:24.385539: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.02291, saving model to lstm_mobV2_gavg_10class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-08 23:59:26.152544: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.153421: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.154016: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:26.265960: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.266617: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.267217: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:26.370718: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.371521: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.372120: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:26.388922: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.393884: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.495790: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.496450: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.497044: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:26.513640: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 23:59:26.610609: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.611450: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.612054: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:26.627076: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 23:59:26.634501: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.640100: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 23:59:26.646034: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.651307: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 23:59:26.656686: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.661254: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.670584: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.675813: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.825989: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-08 23:59:26.836985: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.842166: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:26.938940: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:26.939589: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:26.940184: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:27.045440: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:27.046293: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:27.046915: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:27.078234: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-08 23:59:27.083009: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-08 23:59:27.088028: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:27.092797: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-08 23:59:27.190147: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:27.190813: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:27.191420: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:27.295670: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:27.296558: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:27.297164: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:27.407411: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:27.408090: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:27.408704: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 23:59:27.514175: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 23:59:27.515006: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 23:59:27.515628: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-08 23:59:27.578475: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_10class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_10class_128_unit_4_Adam_ReduceLROnPlateau/assets

2963/2963 [==============================] - 15s 5ms/step - loss: 1.1631 - accuracy: 0.6263 - val_loss: 1.0229 - val_accuracy: 0.6813 - lr: 0.0010
Epoch 2/30
2951/2963 [============================>.] - ETA: 0s - loss: 0.8676 - accuracy: 0.7196
Epoch 2: val_loss improved from 1.02291 to 0.95844, saving model to lstm_mobV2_gavg_10class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(966, 40, 1280)
(966, 10)
242/242 [==============================] - 1s 3ms/step

(966, 10)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[63  2  0 14  2  2  0  5  4  6]
 [ 1 76  0  8  1  1  0  2  2  1]
 [ 3  3 67  7  9  1  1  3  5  1]
 [ 2  8  0 56  7  5  2  4  8  6]
 [ 0  1  5  4 65  4  2  5  5  5]
 [ 0 11  4 21  4 49  0  4  2  1]
 [ 1  0  2 11 11  5 58  2  3  5]
 [ 5  8  7 20 13  4  3 18 10  8]
 [ 2  8  1  7  4  0  0  2 73  3]
 [ 3  2  1 12 10 11  6 10  5 32]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.79      0.64      0.71        98
           1       0.64      0.83      0.72        92
           2       0.77      0.67      0.72       100
           3       0.35      0.57      0.43        98
           4       0.52      0.68      0.59        96
           5       0.60      0.51      0.55        96
           6       0.81      0.59      0.68        98
           7       0.33      0.19      0.24        96
           8       0.62      0.73      0.67       100
           9       0.47      0.35      0.40        92

    accuracy                           0.58       966
   macro avg       0.59      0.58      0.57       966
weighted avg       0.59      0.58      0.57       966

#### Model setup:
Extractor = 'dens201_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1920)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1920)
Shape of y_test:  (2191, 7)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,049,991
Trainable params: 1,049,991
Non-trainable params: 0
_________________________________________________________________

2023-09-08 05:58:06.276581: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:06.277565: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:06.278191: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-08 05:58:10.337009: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.
2023-09-08 05:58:11.800392: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-08 05:58:12.849179: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:12.850229: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:12.850919: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:13.211467: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:13.212487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:13.213155: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:13.585135: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-08 05:58:13.662458: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f3fe4118a10 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-08 05:58:13.662493: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 05:58:13.662498: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-08 05:58:13.691261: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-08 05:58:13.691291: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.

  27/2193 [..............................] - ETA: 8s - loss: 1.9182 - accuracy: 0.2037

2023-09-08 05:58:13.972451: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2187/2193 [============================>.] - ETA: 0s - loss: 0.8636 - accuracy: 0.7042

2023-09-08 05:58:23.050737: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 673075200 exceeds 10% of free system memory.
2023-09-08 05:58:23.418011: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 673075200 exceeds 10% of free system memory.
2023-09-08 05:58:23.766020: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:23.766997: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:23.767600: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.73046, saving model to lstm_dens201_gavg_128_unit_4_Adam_ReduceLROnPlateau

2023-09-08 05:58:25.181404: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.182447: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.183077: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:25.303011: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.303707: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.304372: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:25.415456: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.416511: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.417152: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:25.435124: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.440353: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.548246: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.548933: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.549581: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:25.567285: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-08 05:58:25.670272: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.671224: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.671881: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:25.687943: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-08 05:58:25.695796: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.701715: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-08 05:58:25.707903: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.713407: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-08 05:58:25.719069: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.723779: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.733448: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.738890: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.774276: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-08 05:58:25.785518: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.791153: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:25.893486: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:25.894414: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:25.895066: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:26.130661: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:26.131428: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:26.132261: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:26.164049: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-08 05:58:26.169114: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-08 05:58:26.174431: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:26.179426: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-08 05:58:26.283961: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:26.284647: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:26.285247: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:26.393488: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:26.394394: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:26.395016: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:26.510812: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:26.511501: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:26.512102: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 05:58:26.619480: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 05:58:26.620405: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 05:58:26.621039: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-08 05:58:26.683097: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1920)
(678, 7)
 16/170 [=>............................] - ETA: 0s

2023-09-08 06:04:48.057373: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-08 06:04:48.058136: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-08 06:04:48.058948: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-08 06:04:48.174714: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

170/170 [==============================] - 1s 3ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[70  4  2 10  8  4  0]
 [ 0 82  1  5  4  0  0]
 [ 0  2 56  7 29  3  3]
 [ 3  8  0 61 16 10  0]
 [ 2  0  0  1 87  5  1]
 [ 0  6  0 18  8 62  2]
 [ 1  0  5 14  9  5 64]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.92      0.71      0.80        98
           1       0.80      0.89      0.85        92
           2       0.88      0.56      0.68       100
           3       0.53      0.62      0.57        98
           4       0.54      0.91      0.68        96
           5       0.70      0.65      0.67        96
           6       0.91      0.65      0.76        98

    accuracy                           0.71       678
   macro avg       0.75      0.71      0.72       678
weighted avg       0.75      0.71      0.72       678

#### Model setup:
Extractor = 'dens201_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1920)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1920)
Shape of y_test:  (2480, 8)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 1,050,120
Trainable params: 1,050,120
Non-trainable params: 0
_________________________________________________________________

2023-09-14 00:23:53.146950: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:23:53.147866: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:23:53.148478: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-14 00:23:58.353783: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.
2023-09-14 00:23:59.958602: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.

Epoch 1/30

2023-09-14 00:24:01.068987: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:01.069951: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:01.070600: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:01.428669: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:01.429635: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:01.430277: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:01.786007: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-14 00:24:01.866939: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f33900481f0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-14 00:24:01.866968: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-14 00:24:01.866974: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-14 00:24:01.871627: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-14 00:24:01.986296: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2474/2482 [============================>.] - ETA: 0s - loss: 0.9417 - accuracy: 0.6839

2023-09-14 00:24:12.536610: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:12.537620: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:12.538243: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.77649, saving model to lstm_dens201_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-14 00:24:13.931975: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:13.932884: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:13.933456: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.044994: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.045647: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.046238: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.148992: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.149840: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.150433: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.167355: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.172369: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.395522: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.396444: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.397063: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.414197: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-14 00:24:14.518589: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.519274: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.519872: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.534671: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-14 00:24:14.542387: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.548350: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-14 00:24:14.554520: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.560896: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-14 00:24:14.566592: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.571338: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.580979: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.586574: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.622222: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-14 00:24:14.633644: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.639142: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.738917: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.739610: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.740217: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.846707: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.847603: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.848233: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:14.879183: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-14 00:24:14.884134: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-14 00:24:14.889114: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.893834: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-14 00:24:14.989128: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:14.989748: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:14.990323: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:15.095889: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:15.096789: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:15.097385: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:15.203952: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:15.204610: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:15.205190: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:24:15.306264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:24:15.307215: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:24:15.307821: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-14 00:24:15.370300: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_128_unit_4_Adam_ReduceLROnPlateau/assets

2482/2482 [==============================] - 15s 5ms/step - loss: 0.9403 - accuracy: 0.6844 - val_loss: 0.7765 - val_accuracy: 0.7431 - lr: 0.0010
Epoch 2/30
2476/2482 [============================>.] - ETA: 0s - loss: 0.6390 - accuracy: 0.7900
Epoch 2: val_loss did not improve from 0.77649
2482/2482 [==============================] - 11s 4ms/step - loss: 0.6386 - accuracy: 0.7900 - val_loss: 0.7904 - val_accuracy: 0.7210 - lr: 0.0010
Epoch 3/30
2470/2482 [============================>.] - ETA: 0s - loss: 0.5544 - accuracy: 0.8141
Epoch 3: val_loss did not improve from 0.77649
2482/2482 [==============================] - 11s 4ms/step - loss: 0.5538 - accuracy: 0.8144 - val_loss: 0.8078 - val_accuracy: 0.7153 - lr: 0.0010
Epoch 4/30
2471/2482 [============================>.] - ETA: 0s - loss: 0.5092 - accuracy: 0.8327
Epoch 4: val_loss improved from 0.77649 to 0.75280, saving model to lstm_dens201_gavg8class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1920)
(778, 8)
 18/195 [=>............................] - ETA: 0s

2023-09-14 00:31:17.163880: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:31:17.164557: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:31:17.165338: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-14 00:31:17.288725: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

195/195 [==============================] - 1s 3ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[62  1  1 16  5  8  1  4]
 [ 0 74  2  6  2  5  0  3]
 [ 0  2 68  3  9  4  3 11]
 [ 2  8  1 62  4 15  0  6]
 [ 1  0  0  3 73  5  4 10]
 [ 0  0  2 14  3 75  0  2]
 [ 0  0  3  5  6  9 69  6]
 [ 3  2  2  4  3  3  0 83]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.91      0.63      0.75        98
           1       0.85      0.80      0.83        92
           2       0.86      0.68      0.76       100
           3       0.55      0.63      0.59        98
           4       0.70      0.76      0.73        96
           5       0.60      0.78      0.68        96
           6       0.90      0.70      0.79        98
           7       0.66      0.83      0.74       100

    accuracy                           0.73       778
   macro avg       0.75      0.73      0.73       778
weighted avg       0.75      0.73      0.73       778

#### Model setup:
Extractor = 'dens201_gavg9class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (11078, 40, 1920)
Shape of y_train:  (11078, 9)
Shape of X_test:  (2768, 40, 1920)
Shape of y_test:  (2768, 9)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 9)                 1161      
                                                                 
=================================================================
Total params: 1,050,249
Trainable params: 1,050,249
Non-trainable params: 0
_________________________________________________________________

2023-09-09 04:35:23.459033: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:23.459908: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:23.460585: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-09 04:35:32.382512: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3403161600 exceeds 10% of free system memory.
2023-09-09 04:35:34.311496: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3403161600 exceeds 10% of free system memory.

Epoch 1/30

2023-09-09 04:35:35.594907: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:35.595948: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:35.596654: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:35.971317: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:35.972300: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:35.972972: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:36.350844: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-09 04:35:36.398809: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f57949ed170 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-09 04:35:36.398832: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 04:35:36.398836: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 04:35:36.403988: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-09 04:35:36.509277: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2766/2770 [============================>.] - ETA: 0s - loss: 0.9797 - accuracy: 0.6839

2023-09-09 04:35:47.540041: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 850329600 exceeds 10% of free system memory.
2023-09-09 04:35:48.043198: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 850329600 exceeds 10% of free system memory.
2023-09-09 04:35:48.461900: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:48.462877: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:48.463486: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.81054, saving model to lstm_dens201_gavg9class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-09 04:35:50.110088: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.111000: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.111626: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:50.225632: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.226334: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.226977: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:50.333904: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.334864: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.335495: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:50.352650: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.357837: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.463142: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.463852: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.464495: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:50.481538: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 04:35:50.581349: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.584441: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.585191: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:50.601233: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 04:35:50.609343: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.615530: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 04:35:50.622196: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.628055: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 04:35:50.633997: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.639065: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.649036: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.654862: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.690529: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 04:35:50.701551: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.706985: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:50.927564: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:50.928265: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:50.928985: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:51.037485: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:51.038175: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:51.038803: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:51.070935: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-09 04:35:51.075936: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-09 04:35:51.081114: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:51.085876: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 04:35:51.189722: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:51.190655: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:51.191288: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:51.300089: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:51.300788: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:51.301403: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:51.416374: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:51.417252: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:51.417886: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:35:51.526761: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:35:51.527448: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:35:51.528069: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-09 04:35:51.591101: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg9class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg9class_128_unit_4_Adam_ReduceLROnPlateau/assets

2770/2770 [==============================] - 16s 6ms/step - loss: 0.9792 - accuracy: 0.6841 - val_loss: 0.8105 - val_accuracy: 0.7337 - lr: 0.0010
Epoch 2/30
2764/2770 [============================>.] - ETA: 0s - loss: 0.6672 - accuracy: 0.7845
Epoch 2: val_loss did not improve from 0.81054
2770/2770 [==============================] - 12s 4ms/step - loss: 0.6669 - accuracy: 0.7845 - val_loss: 0.8566 - val_accuracy: 0.7135 - lr: 0.0010
Epoch 3/30
2760/2770 [============================>.] - ETA: 0s - loss: 0.5923 - accuracy: 0.8064
Epoch 3: val_loss improved from 0.81054 to 0.78400, saving model to lstm_dens201_gavg9class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(878, 40, 1920)
(878, 9)

2023-09-09 04:43:16.972160: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 269721600 exceeds 10% of free system memory.

 19/220 [=>............................] - ETA: 0s

2023-09-09 04:43:17.393862: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 04:43:17.394584: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 04:43:17.395267: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 04:43:17.512359: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

220/220 [==============================] - 1s 3ms/step

(878, 9)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[53  1  1 18  7  8  0  8  2]
 [ 0 70  2  5  2  6  0  5  2]
 [ 1  1 56  5 12  8  1 13  3]
 [ 0  3  0 59  4 22  3  4  3]
 [ 0  0  0  6 73  7  1  9  0]
 [ 0  0  0  5  5 81  1  3  1]
 [ 0  1  3  8  6 11 61  7  1]
 [ 2  3  2  2  4  7  0 79  1]
 [ 2  4  1  7  1 15  2  3 65]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.91      0.54      0.68        98
           1       0.84      0.76      0.80        92
           2       0.86      0.56      0.68       100
           3       0.51      0.60      0.55        98
           4       0.64      0.76      0.70        96
           5       0.49      0.84      0.62        96
           6       0.88      0.62      0.73        98
           7       0.60      0.79      0.68       100
           8       0.83      0.65      0.73       100

    accuracy                           0.68       878
   macro avg       0.73      0.68      0.69       878
weighted avg       0.73      0.68      0.69       878

#### Model setup:
Extractor = 'dens201_gavg10class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
data_dir = '/home/ycaun/Desktop/jeheng/Processed dataset/Train_data_processed/'
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (11852, 40, 1920)
Shape of y_train:  (11852, 10)
Shape of X_test:  (2962, 40, 1920)
Shape of y_test:  (2962, 10)

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 10)                1290      
                                                                 
=================================================================
Total params: 1,050,378
Trainable params: 1,050,378
Non-trainable params: 0
_________________________________________________________________

2023-09-09 00:54:27.376418: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:27.377346: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:27.377938: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-09 00:54:28.963395: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3640934400 exceeds 10% of free system memory.
2023-09-09 00:54:30.925803: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3640934400 exceeds 10% of free system memory.

Epoch 1/30

2023-09-09 00:54:32.254555: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:32.256333: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:32.256989: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:32.583434: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:32.584490: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:32.585173: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:32.895361: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-09 00:54:32.937465: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f3adfbc5630 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-09 00:54:32.937487: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 00:54:32.937492: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-09 00:54:32.940839: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-09 00:54:33.045640: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2959/2963 [============================>.] - ETA: 0s - loss: 1.1011 - accuracy: 0.6475

2023-09-09 00:54:44.057581: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 909926400 exceeds 10% of free system memory.
2023-09-09 00:54:44.551623: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 909926400 exceeds 10% of free system memory.
2023-09-09 00:54:44.959471: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:44.960499: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:44.961137: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.95498, saving model to lstm_dens201_gavg10class_128_unit_4_Adam_ReduceLROnPlateau

2023-09-09 00:54:46.810816: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:46.811550: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:46.812313: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:46.925229: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:46.925902: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:46.926500: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.030400: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.031297: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.031939: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.049254: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.054150: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.153848: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.154499: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.155099: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.171302: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 00:54:47.268328: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.269206: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.269822: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.284657: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 00:54:47.292054: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.297541: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 00:54:47.303388: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.308536: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 00:54:47.313762: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.318268: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.327395: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.332494: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.480415: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-09 00:54:47.492435: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.498287: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.600419: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.601124: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.601750: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.711444: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.712390: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.713023: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.744897: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-09 00:54:47.749806: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-09 00:54:47.755136: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.759981: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-09 00:54:47.858716: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.859420: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.860048: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:47.969180: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:47.969973: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:47.970783: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:48.079348: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:48.080006: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:48.080640: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 00:54:48.182884: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 00:54:48.183813: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 00:54:48.184471: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-09 00:54:48.245804: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg10class_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg10class_128_unit_4_Adam_ReduceLROnPlateau/assets

2963/2963 [==============================] - 16s 5ms/step - loss: 1.1005 - accuracy: 0.6477 - val_loss: 0.9550 - val_accuracy: 0.6894 - lr: 0.0010
Epoch 2/30
2950/2963 [============================>.] - ETA: 0s - loss: 0.7797 - accuracy: 0.7556
Epoch 2: val_loss improved from 0.95498 to 0.89108, saving model to lstm_dens201_gavg10class_128_unit_4_Adam_ReduceLROnPlateau

# if X_eval is not None and y_eval is not None:
#     del X_eval, y_eval
#     gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(966, 40, 1920)
(966, 10)
 17/242 [=>............................] - ETA: 0s

2023-09-09 01:02:55.632214: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-09 01:02:55.633123: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-09 01:02:55.633705: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-09 01:02:55.723701: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

242/242 [==============================] - 1s 3ms/step

(966, 10)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[68  0  2 12  2  8  2  0  1  3]
 [ 1 72  2  6  2  6  0  1  2  0]
 [ 0  3 67  6 12  3  4  0  5  0]
 [ 2  2  2 66  1 16  2  4  2  1]
 [ 0  0  1  4 71  6  3  1  9  1]
 [ 0  1  0 17  4 69  2  0  2  1]
 [ 0  0  4 10  6  7 65  1  1  4]
 [ 8  1  5 35  2 14  3 20  5  3]
 [ 5  2  2 14  5  9  1  3 58  1]
 [ 8  0  1 17  3 16  8  4  2 33]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.74      0.69      0.72        98
           1       0.89      0.78      0.83        92
           2       0.78      0.67      0.72       100
           3       0.35      0.67      0.46        98
           4       0.66      0.74      0.70        96
           5       0.45      0.72      0.55        96
           6       0.72      0.66      0.69        98
           7       0.59      0.21      0.31        96
           8       0.67      0.58      0.62       100
           9       0.70      0.36      0.47        92

    accuracy                           0.61       966
   macro avg       0.65      0.61      0.61       966
weighted avg       0.65      0.61      0.61       966

#### Model setup:
Extractor = 'resnet_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('resnet_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30
546/549 [============================>.] - ETA: 0s - loss: 1.8456 - accuracy: 0.2382
Epoch 1: val_loss improved from inf to 1.76665, saving model to lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 15s 18ms/step - loss: 1.8460 - accuracy: 0.2381 - val_loss: 1.7666 - val_accuracy: 0.3008 - lr: 0.0010
Epoch 2/30
544/549 [============================>.] - ETA: 0s - loss: 1.7339 - accuracy: 0.2993
Epoch 2: val_loss improved from 1.76665 to 1.70040, saving model to lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 6s 12ms/step - loss: 1.7339 - accuracy: 0.2992 - val_loss: 1.7004 - val_accuracy: 0.3140 - lr: 0.0010
Epoch 3/30
548/549 [============================>.] - ETA: 0s - loss: 1.6848 - accuracy: 0.3341
Epoch 3: val_loss improved from 1.70040 to 1.64049, saving model to lstm_resnet_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('resnet_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 2s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[29  1  1 32 22 11  2]
 [ 4 45  6 11  0 26  0]
 [ 6  6 27 21 11 18 11]
 [ 8  2 12 29 20 21  6]
 [15  3  6 19 34 10  9]
 [ 6  5  4 21  1 55  4]
 [ 8  0 15 19  9 11 36]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.38      0.30      0.33        98
           1       0.73      0.49      0.58        92
           2       0.38      0.27      0.32       100
           3       0.19      0.30      0.23        98
           4       0.35      0.35      0.35        96
           5       0.36      0.57      0.44        96
           6       0.53      0.37      0.43        98

    accuracy                           0.38       678
   macro avg       0.42      0.38      0.39       678
weighted avg       0.41      0.38      0.38       678

#### Model setup:
Extractor = 'vgg16_avg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_avg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30
547/549 [============================>.] - ETA: 0s - loss: 1.1509 - accuracy: 0.5882
Epoch 1: val_loss improved from inf to 1.03062, saving model to lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 17s 15ms/step - loss: 1.1509 - accuracy: 0.5879 - val_loss: 1.0306 - val_accuracy: 0.6276 - lr: 0.0010
Epoch 2/30
545/549 [============================>.] - ETA: 0s - loss: 0.8823 - accuracy: 0.6945
Epoch 2: val_loss improved from 1.03062 to 0.97363, saving model to lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 7s 12ms/step - loss: 0.8822 - accuracy: 0.6945 - val_loss: 0.9736 - val_accuracy: 0.6486 - lr: 0.0010
Epoch 3/30
547/549 [============================>.] - ETA: 0s - loss: 0.7485 - accuracy: 0.7401
Epoch 3: val_loss improved from 0.97363 to 0.87443, saving model to lstm_vgg16_avg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_avg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 4s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[66  2  3 14  1  5  7]
 [ 1 66  7  5  7  4  2]
 [ 9  4 72  1  8  0  6]
 [ 9 11  6 55  4  8  5]
 [ 7  2 11 12 57  2  5]
 [ 0  1  6 19  3 63  4]
 [ 7  2  9  8  7  4 61]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.67      0.67      0.67        98
           1       0.75      0.72      0.73        92
           2       0.63      0.72      0.67       100
           3       0.48      0.56      0.52        98
           4       0.66      0.59      0.62        96
           5       0.73      0.66      0.69        96
           6       0.68      0.62      0.65        98

    accuracy                           0.65       678
   macro avg       0.66      0.65      0.65       678
weighted avg       0.66      0.65      0.65       678

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8768, 40, 512)
Shape of y_train:  (8768, 7)
Shape of X_test:  (2192, 40, 512)
Shape of y_test:  (2192, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 128)               328192    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 329,095
Trainable params: 329,095
Non-trainable params: 0
_________________________________________________________________

79

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30
541/548 [============================>.] - ETA: 0s - loss: 1.2121 - accuracy: 0.5587
Epoch 1: val_loss improved from inf to 1.17019, saving model to lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

548/548 [==============================] - 24s 10ms/step - loss: 1.2090 - accuracy: 0.5594 - val_loss: 1.1702 - val_accuracy: 0.5867 - lr: 0.0010
Epoch 2/30
544/548 [============================>.] - ETA: 0s - loss: 0.9480 - accuracy: 0.6667
Epoch 2: val_loss improved from 1.17019 to 0.97771, saving model to lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

548/548 [==============================] - 5s 9ms/step - loss: 0.9471 - accuracy: 0.6672 - val_loss: 0.9777 - val_accuracy: 0.6492 - lr: 0.0010
Epoch 3/30
547/548 [============================>.] - ETA: 0s - loss: 0.8290 - accuracy: 0.7070
Epoch 3: val_loss improved from 0.97771 to 0.86261, saving model to lstm_vgg16_gavg_128_unit_16_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 512)
(678, 7)
170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[67  0  2 18  2  4  5]
 [ 2 72  5  4  5  3  1]
 [ 9  7 67  3  8  0  6]
 [15  8  3 51  7  8  6]
 [ 3  3  9  8 64  4  5]
 [ 2  7  7 14  6 55  5]
 [ 6  2  9  8  5  3 65]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.64      0.68      0.66        98
           1       0.73      0.78      0.75        92
           2       0.66      0.67      0.66       100
           3       0.48      0.52      0.50        98
           4       0.66      0.67      0.66        96
           5       0.71      0.57      0.64        96
           6       0.70      0.66      0.68        98

    accuracy                           0.65       678
   macro avg       0.65      0.65      0.65       678
weighted avg       0.65      0.65      0.65       678

#### Model setup:
Extractor = 'mobV2_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'
NUM_CLASSES = 7

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1280)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1280)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 722,311
Trainable params: 722,311
Non-trainable params: 0
_________________________________________________________________

2023-09-12 02:51:07.366022: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:07.367082: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:07.367706: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

77

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30

2023-09-12 02:51:09.143131: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:09.144160: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:09.144863: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:09.500783: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:09.501965: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:09.502635: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

544/549 [============================>.] - ETA: 0s - loss: 0.8728 - accuracy: 0.7021

2023-09-12 02:51:13.205598: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:13.206581: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:13.207195: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.77459, saving model to lstm_mobV2_gavg_128_unit_16_Adam_ReduceLROnPlateau

2023-09-12 02:51:13.828950: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:13.829955: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:13.830559: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:13.946586: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:13.947302: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:13.947915: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.054627: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.055582: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.056203: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.073327: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.078282: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.183280: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.184011: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.184674: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.201979: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:14.301945: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.302944: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.303586: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.318535: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:14.326015: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.331467: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:14.337116: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.342283: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:14.347676: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.352173: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.361470: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.366615: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.382298: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:14.392574: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.397842: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.492911: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.493586: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.494197: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.597425: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.598447: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.599069: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.628717: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 02:51:14.633204: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 02:51:14.637972: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.642530: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:14.738253: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.738898: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.739508: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.839463: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.840476: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.841070: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:14.946785: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:14.947452: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:14.948043: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:15.046748: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:15.047658: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:15.048271: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 02:51:15.105746: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 6s 10ms/step - loss: 0.8722 - accuracy: 0.7024 - val_loss: 0.7746 - val_accuracy: 0.7408 - lr: 0.0010
Epoch 2/30
540/549 [============================>.] - ETA: 0s - loss: 0.6579 - accuracy: 0.7801
Epoch 2: val_loss improved from 0.77459 to 0.74453, saving model to lstm_mobV2_gavg_128_unit_16_Adam_ReduceLROnPlateau

2023-09-12 02:51:18.479196: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:18.480141: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:18.480915: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:18.600386: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:18.601096: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:18.601729: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:18.712209: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:18.713140: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:18.713776: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:18.733015: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:18.738287: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:18.847184: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:18.847881: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:18.848525: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:18.866219: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:18.966308: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:18.967220: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:18.967839: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:18.983091: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:18.991312: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:18.997198: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:19.003264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.008485: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:19.260593: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.267045: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.277403: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.283078: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.299945: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 02:51:19.311109: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.317116: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.419273: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:19.419951: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:19.420577: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:19.531201: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:19.532191: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:19.532812: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:19.557211: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 02:51:19.562068: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 02:51:19.567248: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.572033: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 02:51:19.673180: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:19.673851: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:19.674497: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:19.781874: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:19.782850: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:19.783485: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:19.896115: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:19.896801: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:19.897442: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 02:51:20.005458: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:51:20.006402: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:51:20.007043: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 02:51:20.067565: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1280)
(678, 7)
 74/170 [============>.................] - ETA: 0s

2023-09-12 02:54:58.355537: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 02:54:58.356462: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 02:54:58.357099: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[80  1  3  9  3  2  0]
 [ 1 74  4  8  1  4  0]
 [ 2  3 72  1 15  2  5]
 [ 5  9  6 59  7 10  2]
 [ 4  1  8  1 77  3  2]
 [ 1  4  2 16  7 63  3]
 [ 3  0  5  7 11  7 65]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.83      0.82      0.82        98
           1       0.80      0.80      0.80        92
           2       0.72      0.72      0.72       100
           3       0.58      0.60      0.59        98
           4       0.64      0.80      0.71        96
           5       0.69      0.66      0.67        96
           6       0.84      0.66      0.74        98

    accuracy                           0.72       678
   macro avg       0.73      0.72      0.72       678
weighted avg       0.73      0.72      0.72       678

#### Model setup:
Extractor = 'dens201_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'
NUM_CLASSES = 7

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1920)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1920)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,049,991
Trainable params: 1,049,991
Non-trainable params: 0
_________________________________________________________________

2023-09-12 04:27:17.845607: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:17.846502: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:17.847090: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

77

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-12 04:27:19.219945: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 04:27:21.568273: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:21.569229: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:21.569911: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:21.899273: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:21.900211: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:21.900843: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

542/549 [============================>.] - ETA: 0s - loss: 0.8567 - accuracy: 0.7107

2023-09-12 04:27:26.119682: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:26.120626: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:26.121226: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.74743, saving model to lstm_dens201_gavg_128_unit_16_Adam_ReduceLROnPlateau

2023-09-12 04:27:26.821949: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:26.822628: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:26.823332: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:26.930270: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:26.930902: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:26.931503: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.029964: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.030776: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.031362: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.048230: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.052970: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.151957: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.152608: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.153193: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.169470: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:27:27.265572: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.266457: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.267051: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.282149: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:27:27.289476: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.294857: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:27:27.300562: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.305689: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:27:27.310878: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.315281: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.324486: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.329657: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.345126: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:27:27.355531: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.360857: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.457471: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.458137: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.458722: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.565909: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.566710: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.567311: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.597114: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 04:27:27.601804: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 04:27:27.606613: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.611029: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:27:27.706444: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.707122: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.707747: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.811118: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.811811: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.812550: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:27.915605: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:27.916241: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:27.916823: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:27:28.014012: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:27:28.014846: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:27:28.015448: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 04:27:28.075610: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg_128_unit_16_Adam_ReduceLROnPlateau/assets

549/549 [==============================] - 7s 11ms/step - loss: 0.8562 - accuracy: 0.7106 - val_loss: 0.7474 - val_accuracy: 0.7549 - lr: 0.0010
Epoch 2/30
547/549 [============================>.] - ETA: 0s - loss: 0.5862 - accuracy: 0.8029
Epoch 2: val_loss improved from 0.74743 to 0.68359, saving model to lstm_dens201_gavg_128_unit_16_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1920)
(678, 7)
 64/170 [==========>...................] - ETA: 0s

2023-09-12 04:29:51.382484: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:29:51.383486: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:29:51.384116: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[65  1  6 14  6  3  3]
 [ 0 81  3  6  0  2  0]
 [ 1  2 80  8  6  3  0]
 [ 3  5  4 73  3  8  2]
 [ 3  1  9  3 72  4  4]
 [ 0  1  2 20  6 63  4]
 [ 0  0  6  8  5  3 76]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.90      0.66      0.76        98
           1       0.89      0.88      0.89        92
           2       0.73      0.80      0.76       100
           3       0.55      0.74      0.63        98
           4       0.73      0.75      0.74        96
           5       0.73      0.66      0.69        96
           6       0.85      0.78      0.81        98

    accuracy                           0.75       678
   macro avg       0.77      0.75      0.76       678
weighted avg       0.77      0.75      0.76       678

#### Model setup:
Extractor = 'mobV2_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1280)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1280)
Shape of y_test:  (2480, 8)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 722,440
Trainable params: 722,440
Non-trainable params: 0
_________________________________________________________________

2023-09-12 05:55:19.154948: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:19.155834: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:19.156450: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

77

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-09-12 05:55:22.059296: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.
2023-09-12 05:55:23.174808: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 05:55:24.011787: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:24.012726: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:24.013391: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:24.358568: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:24.359593: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:24.360253: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

614/621 [============================>.] - ETA: 0s - loss: 0.9613 - accuracy: 0.6857

2023-09-12 05:55:28.662685: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:28.663656: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:28.664271: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.89010, saving model to lstm_mobV2_gavg8class_128_unit_16_Adam_ReduceLROnPlateau

2023-09-12 05:55:29.296439: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.297365: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.297970: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:29.412368: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.413055: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.413659: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:29.513478: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.514317: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.514924: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:29.530843: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.535572: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.632204: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.632842: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.633445: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:29.649059: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:55:29.741380: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.742187: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.742753: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:29.757202: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:55:29.764252: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.769587: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:55:29.775169: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.780210: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:55:29.785227: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.789452: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.798142: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.803075: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.818012: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:55:29.827810: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.832936: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:29.927359: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:29.927986: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:29.928580: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:30.030565: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:30.031438: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:30.032034: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:30.060351: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:55:30.064793: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:55:30.069512: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:30.073904: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:55:30.165803: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:30.166474: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:30.167075: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:30.266426: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:30.267263: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:30.267861: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:30.374349: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:30.375004: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:30.375612: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:55:30.477984: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:55:30.478807: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:55:30.479421: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 05:55:30.538503: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_16_Adam_ReduceLROnPlateau/assets

621/621 [==============================] - 7s 9ms/step - loss: 0.9595 - accuracy: 0.6861 - val_loss: 0.8901 - val_accuracy: 0.7109 - lr: 0.0010
Epoch 2/30
612/621 [============================>.] - ETA: 0s - loss: 0.6638 - accuracy: 0.7822
Epoch 2: val_loss improved from 0.89010 to 0.78298, saving model to lstm_mobV2_gavg8class_128_unit_16_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1280)
(778, 8)
 56/195 [=======>......................] - ETA: 0s

2023-09-12 05:58:17.983749: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:58:17.984808: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:58:17.985454: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 1s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[73  0  2  9  3  3  3  5]
 [ 1 77  3  4  3  3  0  1]
 [ 2  2 63  2 18  3  2  8]
 [ 5 10  2 56 10  5  1  9]
 [ 3  4  5  2 74  2  2  4]
 [ 0  3  3 12  4 66  6  2]
 [ 3  2  4  8  7  4 66  4]
 [ 1  5  4  7  4  5  1 73]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.83      0.74      0.78        98
           1       0.75      0.84      0.79        92
           2       0.73      0.63      0.68       100
           3       0.56      0.57      0.57        98
           4       0.60      0.77      0.68        96
           5       0.73      0.69      0.71        96
           6       0.81      0.67      0.74        98
           7       0.69      0.73      0.71       100

    accuracy                           0.70       778
   macro avg       0.71      0.71      0.71       778
weighted avg       0.71      0.70      0.71       778

#### Model setup:
Extractor = 'dens201_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '16'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1920)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1920)
Shape of y_test:  (2480, 8)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 1,050,120
Trainable params: 1,050,120
Non-trainable params: 0
_________________________________________________________________

2023-09-13 23:37:42.358078: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:42.358778: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:42.359476: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

77

history = model.fit(X_train, y_train, batch_size=16, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

Epoch 1/30

2023-09-13 23:37:48.990737: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:48.991800: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:48.992456: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:49.335704: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:49.336662: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:49.337312: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

615/621 [============================>.] - ETA: 0s - loss: 0.9364 - accuracy: 0.6864

2023-09-13 23:37:54.101692: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:54.102434: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:54.103205: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.79470, saving model to lstm_dens201_gavg8class_128_unit_16_Adam_ReduceLROnPlateau

2023-09-13 23:37:54.842393: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:54.843087: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:54.843801: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:54.951027: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:54.951663: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:54.952264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.051106: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.051892: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.052487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.068344: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.073051: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.169394: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.170040: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.170629: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.186245: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:37:55.278773: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.279563: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.280176: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.294536: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:37:55.301516: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.306807: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:37:55.312370: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.317399: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:37:55.322441: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.326720: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.335458: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.340358: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.355284: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:37:55.365067: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.370106: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.463507: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.464145: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.464761: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.567142: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.568022: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.568655: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.597550: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:37:55.602028: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:37:55.606696: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.611135: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:37:55.703259: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.703879: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.704466: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.803268: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.804085: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.804695: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:55.908213: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:55.908859: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:55.909463: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:37:56.007606: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:37:56.008418: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:37:56.009022: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-13 23:37:56.066273: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_128_unit_16_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_128_unit_16_Adam_ReduceLROnPlateau/assets

621/621 [==============================] - 7s 10ms/step - loss: 0.9343 - accuracy: 0.6876 - val_loss: 0.7947 - val_accuracy: 0.7391 - lr: 0.0010
Epoch 2/30
615/621 [============================>.] - ETA: 0s - loss: 0.6413 - accuracy: 0.7894
Epoch 2: val_loss improved from 0.79470 to 0.75407, saving model to lstm_dens201_gavg8class_128_unit_16_Adam_ReduceLROnPlateau

2023-09-13 23:38:00.539264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:00.541216: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:00.541883: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:00.658934: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:00.659611: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:00.660229: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:00.955609: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:00.956557: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:00.957196: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:00.974053: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:00.979146: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.081553: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.082304: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.082926: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.099594: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:38:01.197209: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.198177: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.198772: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.213744: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:38:01.221085: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.226704: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:38:01.232623: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.237878: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:38:01.243164: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.247598: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.256732: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.261765: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.277288: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:38:01.287872: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.293140: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.393053: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.393733: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.394374: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.504439: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.505491: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.506130: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.531361: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:38:01.536370: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:38:01.541297: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.545806: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:38:01.643785: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.644493: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.645122: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.748712: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.749566: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.750189: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.858505: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.859169: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.859774: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:38:01.963730: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:38:01.964552: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:38:01.965144: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-13 23:38:02.024466: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1920)
(778, 8)
 68/195 [=========>....................] - ETA: 0s

2023-09-13 23:40:34.068025: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:40:34.068743: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:40:34.069507: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 0s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[65  2  0 21  4  3  2  1]
 [ 0 82  2  5  1  1  0  1]
 [ 1  2 70  5  7  3  5  7]
 [ 3  4  0 68  4 11  2  6]
 [ 1  0  1  6 73  3  4  8]
 [ 0  4  0 13  4 68  3  4]
 [ 0  0  4  8  5  4 74  3]
 [ 4  3  1  8  6  1  1 76]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.88      0.66      0.76        98
           1       0.85      0.89      0.87        92
           2       0.90      0.70      0.79       100
           3       0.51      0.69      0.59        98
           4       0.70      0.76      0.73        96
           5       0.72      0.71      0.72        96
           6       0.81      0.76      0.78        98
           7       0.72      0.76      0.74       100

    accuracy                           0.74       778
   macro avg       0.76      0.74      0.75       778
weighted avg       0.76      0.74      0.74       778

#### Model setup:
Extractor = 'resnet_gavg'
Layer = '128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('resnet_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.001)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

3966

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-04-18 15:32:56.034065: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.
2023-04-18 15:32:58.236971: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.

Epoch 1/30

2023-04-18 15:33:24.922508: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:428] Loaded cuDNN version 8401

2193/2193 [==============================] - ETA: 0s - loss: 1.8237 - accuracy: 0.2518

2023-04-18 15:33:40.473614: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.
2023-04-18 15:33:40.970999: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.

Epoch 1: val_loss improved from inf to 1.71704, saving model to lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 45s 6ms/step - loss: 1.8237 - accuracy: 0.2518 - val_loss: 1.7170 - val_accuracy: 0.3437 - lr: 0.0010
Epoch 2/30
2189/2193 [============================>.] - ETA: 0s - loss: 1.6916 - accuracy: 0.3275
Epoch 2: val_loss improved from 1.71704 to 1.65209, saving model to lstm_resnet_gavg_128_unit_4_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('resnet_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)

2023-04-18 15:39:31.815893: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 222167040 exceeds 10% of free system memory.

170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[26  1  5 20 28 15  3]
 [ 3 57 12  7  1 12  0]
 [ 2  8 40  9 19 15  7]
 [ 6  6 14 25 21 22  4]
 [11  4 10 12 36 14  9]
 [ 3  9 13 22  1 46  2]
 [ 8  3 23  9 12 14 29]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.44      0.27      0.33        98
           1       0.65      0.62      0.63        92
           2       0.34      0.40      0.37       100
           3       0.24      0.26      0.25        98
           4       0.31      0.38      0.34        96
           5       0.33      0.48      0.39        96
           6       0.54      0.30      0.38        98

    accuracy                           0.38       678
   macro avg       0.41      0.38      0.38       678
weighted avg       0.40      0.38      0.38       678

#### Model setup:
Extractor = 'vgg16_avg'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_avg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1114624   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,115,527
Trainable params: 1,115,527
Non-trainable params: 0
_________________________________________________________________

3948

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30
2193/2193 [==============================] - ETA: 0s - loss: 1.0849 - accuracy: 0.6049
Epoch 1: val_loss improved from inf to 1.03485, saving model to lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2193/2193 [==============================] - 14s 6ms/step - loss: 1.0849 - accuracy: 0.6049 - val_loss: 1.0349 - val_accuracy: 0.6335 - lr: 0.0010
Epoch 2/30
2190/2193 [============================>.] - ETA: 0s - loss: 0.8285 - accuracy: 0.7084
Epoch 2: val_loss improved from 1.03485 to 0.97460, saving model to lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2193/2193 [==============================] - 12s 5ms/step - loss: 0.8289 - accuracy: 0.7083 - val_loss: 0.9746 - val_accuracy: 0.6641 - lr: 0.0010
Epoch 3/30
2189/2193 [============================>.] - ETA: 0s - loss: 0.7649 - accuracy: 0.7353
Epoch 3: val_loss improved from 0.97460 to 0.95008, saving model to lstm_vgg16_avg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_avg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[62  2  2 19  3  5  5]
 [ 3 69  7  8  0  5  0]
 [ 8  5 71  3  6  2  5]
 [12  6 12 53  2  3 10]
 [ 7  4 15 10 51  3  6]
 [ 3  2 10 16  1 61  3]
 [ 8  0  9 10  3  9 59]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.60      0.63      0.62        98
           1       0.78      0.75      0.77        92
           2       0.56      0.71      0.63       100
           3       0.45      0.54      0.49        98
           4       0.77      0.53      0.63        96
           5       0.69      0.64      0.66        96
           6       0.67      0.60      0.63        98

    accuracy                           0.63       678
   macro avg       0.65      0.63      0.63       678
weighted avg       0.65      0.63      0.63       678

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8768, 40, 512)
Shape of y_train:  (8768, 7)
Shape of X_test:  (2192, 40, 512)
Shape of y_test:  (2192, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 128)               328192    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 329,095
Trainable params: 329,095
Non-trainable params: 0
_________________________________________________________________

228

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30
2180/2192 [============================>.] - ETA: 0s - loss: 1.1522 - accuracy: 0.5835
Epoch 1: val_loss improved from inf to 1.04646, saving model to lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2192/2192 [==============================] - 10s 4ms/step - loss: 1.1521 - accuracy: 0.5833 - val_loss: 1.0465 - val_accuracy: 0.6268 - lr: 0.0010
Epoch 2/30
2175/2192 [============================>.] - ETA: 0s - loss: 0.9345 - accuracy: 0.6678
Epoch 2: val_loss improved from 1.04646 to 1.02223, saving model to lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2192/2192 [==============================] - 9s 4ms/step - loss: 0.9346 - accuracy: 0.6679 - val_loss: 1.0222 - val_accuracy: 0.6382 - lr: 0.0010
Epoch 3/30
2182/2192 [============================>.] - ETA: 0s - loss: 0.8836 - accuracy: 0.6877
Epoch 3: val_loss did not improve from 1.02223
2192/2192 [==============================] - 7s 3ms/step - loss: 0.8838 - accuracy: 0.6877 - val_loss: 1.0312 - val_accuracy: 0.6401 - lr: 0.0010
Epoch 4/30
2189/2192 [============================>.] - ETA: 0s - loss: 0.8631 - accuracy: 0.6945
Epoch 4: val_loss improved from 1.02223 to 1.01383, saving model to lstm_vgg16_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 512)
(678, 7)
170/170 [==============================] - 1s 1ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[79  1  2 13  1  0  2]
 [ 7 65  8  9  0  1  2]
 [ 9  7 72  3  3  0  6]
 [16  4  8 62  3  0  5]
 [18  5 15  8 38  0 12]
 [13  3 12 28  2 35  3]
 [12  0  8 12  1  4 61]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.51      0.81      0.63        98
           1       0.76      0.71      0.73        92
           2       0.58      0.72      0.64       100
           3       0.46      0.63      0.53        98
           4       0.79      0.40      0.53        96
           5       0.88      0.36      0.51        96
           6       0.67      0.62      0.65        98

    accuracy                           0.61       678
   macro avg       0.66      0.61      0.60       678
weighted avg       0.66      0.61      0.60       678

#### Model setup:
Extractor = 'mobV2_gavg'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1280)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1280)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 722,311
Trainable params: 722,311
Non-trainable params: 0
_________________________________________________________________

2023-09-12 04:36:53.558340: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:36:53.559253: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:36:53.559861: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

77

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30

2023-09-12 04:37:21.855744: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:21.856488: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:21.857290: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:22.145224: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:22.146159: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:22.146857: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

2185/2193 [============================>.] - ETA: 0s - loss: 0.8135 - accuracy: 0.7167

2023-09-12 04:37:31.375127: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:31.376180: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:31.376808: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.72634, saving model to lstm_mobV2_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

2023-09-12 04:37:32.735102: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:32.735942: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:32.736549: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:32.850834: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:32.851521: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:32.852146: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:32.958127: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:32.959018: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:32.959666: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:32.976378: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:32.981317: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.082430: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.083115: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.083726: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.099984: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:33.196738: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.197604: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.198230: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.212873: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:33.220405: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.225892: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:33.231672: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.236872: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:33.242118: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.246526: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.255689: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.261006: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.276875: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:33.287154: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.292425: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.395075: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.395769: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.396390: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.505590: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.506447: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.507061: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.529965: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 04:37:33.534754: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 04:37:33.539607: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.544199: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:33.645417: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.646098: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.646745: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.752007: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.752906: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.753503: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.862414: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.863089: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.863732: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:33.969312: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:33.970203: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:33.970842: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 04:37:34.030351: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2193/2193 [==============================] - 13s 5ms/step - loss: 0.8136 - accuracy: 0.7168 - val_loss: 0.7263 - val_accuracy: 0.7558 - lr: 0.0010
Epoch 2/30
2191/2193 [============================>.] - ETA: 0s - loss: 0.5868 - accuracy: 0.7964
Epoch 2: val_loss improved from 0.72634 to 0.69048, saving model to lstm_mobV2_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

2023-09-12 04:37:43.994555: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:43.996400: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:43.997004: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.106308: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.106935: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.107519: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.207722: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.208493: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.209072: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.225045: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.229768: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.328810: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.329446: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.330030: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.345820: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:44.440239: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.440923: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.441624: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.456111: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:44.463236: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.468574: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:44.474353: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.479456: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:44.485178: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.489507: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.498569: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.503693: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.519015: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 04:37:44.528928: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.534004: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.629707: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.630349: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.630951: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.948508: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:44.949389: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:44.949963: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:44.971574: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 04:37:44.976188: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 04:37:44.981023: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:44.985373: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 04:37:45.079637: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:45.080308: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:45.080895: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:45.182033: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:45.182857: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:45.183449: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:45.291346: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:45.291993: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:45.292589: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:37:45.395863: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:37:45.396668: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:37:45.397390: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 04:37:45.457386: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1280)
(678, 7)
 80/170 [=============>................] - ETA: 0s

2023-09-12 04:44:53.740299: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:44:53.741266: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:44:53.741897: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[78  3  2  9  3  1  2]
 [ 1 72  4  5  4  6  0]
 [ 5  3 79  3  6  3  1]
 [10  7  8 53  4  7  9]
 [ 6  2  7  5 69  3  4]
 [ 3  6  5 13  3 65  1]
 [ 0  2  4  7  7  7 71]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.76      0.80      0.78        98
           1       0.76      0.78      0.77        92
           2       0.72      0.79      0.76       100
           3       0.56      0.54      0.55        98
           4       0.72      0.72      0.72        96
           5       0.71      0.68      0.69        96
           6       0.81      0.72      0.76        98

    accuracy                           0.72       678
   macro avg       0.72      0.72      0.72       678
weighted avg       0.72      0.72      0.72       678

#### Model setup:
Extractor = 'dens201_gavg'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1920)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1920)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 1,049,991
Trainable params: 1,049,991
Non-trainable params: 0
_________________________________________________________________

2023-09-12 04:54:44.542309: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:54:44.543232: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:54:44.543849: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

125

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-12 04:54:51.575824: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.
2023-09-12 04:54:53.046644: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 04:54:54.121826: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:54:54.122875: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:54:54.123550: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:54:54.418142: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:54:54.419190: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:54:54.419853: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

  27/2193 [..............................] - ETA: 8s - loss: 1.8909 - accuracy: 0.3241

2023-09-12 04:54:54.782360: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

2192/2193 [============================>.] - ETA: 0s - loss: 0.8181 - accuracy: 0.7165

2023-09-12 04:55:04.161582: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:04.162324: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:04.163171: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.70026, saving model to lstm_dens201_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau

2023-09-12 04:55:05.580511: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:05.581500: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:05.582118: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:05.699252: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:05.699930: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:05.700548: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:05.806757: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:05.807684: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:05.808307: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:05.825666: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:05.830639: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:05.937702: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:05.938400: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:05.939021: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:05.956394: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:55:06.058345: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.059374: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.060029: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.075940: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:55:06.083879: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.090038: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:55:06.096446: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.101947: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:55:06.107573: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.112236: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.122085: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.127654: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.163134: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 04:55:06.174479: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.180127: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.283424: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.284524: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.285172: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.392082: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.392764: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.393382: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.415313: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 04:55:06.420099: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 04:55:06.425244: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.429930: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 04:55:06.525874: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.526610: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.527385: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.628092: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.628745: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.629348: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.854625: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.855602: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.856228: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 04:55:06.956492: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 04:55:06.957157: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 04:55:06.957749: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 04:55:07.018601: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1920)
(678, 7)
 73/170 [===========>..................] - ETA: 0s

2023-09-12 05:01:09.794612: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:01:09.795480: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:01:09.796128: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[76  0  2 14  4  2  0]
 [ 0 76  2  8  1  5  0]
 [ 2  2 76  2  9  2  7]
 [ 7  7  2 63  8  9  2]
 [ 4  0  5  2 81  2  2]
 [ 2  1  3 14  6 66  4]
 [ 2  1  7  8  5  6 69]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.82      0.78      0.80        98
           1       0.87      0.83      0.85        92
           2       0.78      0.76      0.77       100
           3       0.57      0.64      0.60        98
           4       0.71      0.84      0.77        96
           5       0.72      0.69      0.70        96
           6       0.82      0.70      0.76        98

    accuracy                           0.75       678
   macro avg       0.76      0.75      0.75       678
weighted avg       0.76      0.75      0.75       678

#### Model setup:
Extractor = 'mobV2_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'mobV2_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1280)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1280)
Shape of y_test:  (2480, 8)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               721408    
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 722,440
Trainable params: 722,440
Non-trainable params: 0
_________________________________________________________________

2023-09-12 05:06:45.318298: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:06:45.319210: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:06:45.319807: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

3840

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-12 05:06:48.833804: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 05:06:50.608540: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:06:50.609629: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:06:50.610284: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:06:50.908453: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:06:50.909573: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:06:50.910239: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

2474/2482 [============================>.] - ETA: 0s - loss: 0.9103 - accuracy: 0.6920

2023-09-12 05:07:01.030992: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:01.031685: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:01.032499: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.80617, saving model to lstm_mobV2_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau

2023-09-12 05:07:02.387052: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:02.388049: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:02.388702: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:02.504269: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:02.504953: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:02.505590: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:02.609085: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:02.610158: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:02.610785: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:02.627827: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.632888: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.737107: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:02.737787: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:02.738418: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:02.755273: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:07:02.853920: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:02.854944: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:02.855596: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:02.870764: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:07:02.878285: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.883873: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:07:02.889802: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.895126: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:07:02.900385: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.904776: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.913858: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.918912: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.934690: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:07:02.945119: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:02.950516: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:03.048932: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.049607: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.050217: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:03.155756: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.156731: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.157335: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:03.179442: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:07:03.184370: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:07:03.189317: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:03.193924: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:07:03.293269: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.293962: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.294592: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:03.397816: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.398817: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.399440: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:03.510298: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.510974: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.511633: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:07:03.617091: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:07:03.618222: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:07:03.618909: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 05:07:03.677697: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

2482/2482 [==============================] - 13s 5ms/step - loss: 0.9096 - accuracy: 0.6922 - val_loss: 0.8062 - val_accuracy: 0.7294 - lr: 0.0010
Epoch 2/30
2472/2482 [============================>.] - ETA: 0s - loss: 0.6623 - accuracy: 0.7815
Epoch 2: val_loss improved from 0.80617 to 0.78350, saving model to lstm_mobV2_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1280)
(778, 8)
 73/195 [==========>...................] - ETA: 0s

2023-09-12 05:13:38.761908: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:13:38.762848: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:13:38.763460: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 0s 1ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[74  0  3 12  2  1  3  3]
 [ 0 77  3  8  1  3  0  0]
 [ 6  2 75  0  5  2  4  6]
 [ 9  8  2 58  5  2  6  8]
 [ 3  2  4  3 70  4  3  7]
 [ 1  4  3 15  6 64  1  2]
 [ 2  0  4 10  5  4 70  3]
 [ 2  5  6  4  5  3  3 72]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.76      0.76      0.76        98
           1       0.79      0.84      0.81        92
           2       0.75      0.75      0.75       100
           3       0.53      0.59      0.56        98
           4       0.71      0.73      0.72        96
           5       0.77      0.67      0.72        96
           6       0.78      0.71      0.74        98
           7       0.71      0.72      0.72       100

    accuracy                           0.72       778
   macro avg       0.72      0.72      0.72       778
weighted avg       0.72      0.72      0.72       778

#### Model setup:
Extractor = 'dens201_gavg8class'
Layer = '128_unit'
Optimizer = 'Adam_with_decay'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir, 'dens201_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1920)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1920)
Shape of y_test:  (2480, 8)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(128, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

adam = Adam(decay = 0.01)
mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer=adam ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 128)               1049088   
                                                                 
 dense (Dense)               (None, 8)                 1032      
                                                                 
=================================================================
Total params: 1,050,120
Trainable params: 1,050,120
Non-trainable params: 0
_________________________________________________________________

2023-09-12 05:21:18.421324: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:18.422245: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:18.422866: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

0

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-12 05:21:21.558121: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.
2023-09-12 05:21:23.231928: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 05:21:24.401487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:24.402546: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:24.403204: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:24.700127: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:24.701156: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:24.701801: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

  25/2482 [..............................] - ETA: 10s - loss: 2.2466 - accuracy: 0.1400

2023-09-12 05:21:25.064649: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

2472/2482 [============================>.] - ETA: 0s - loss: 0.9080 - accuracy: 0.6967

2023-09-12 05:21:35.779900: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:35.780937: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:35.781578: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 0.80987, saving model to lstm_dens201_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau

2023-09-12 05:21:37.337572: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:37.338654: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:37.339367: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:37.458498: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:37.459181: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:37.459811: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:37.571688: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:37.573226: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:37.573877: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:37.591533: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.597103: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.699358: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:37.700010: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:37.700612: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:37.718322: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:21:37.819415: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:37.820144: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:37.820905: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:37.836227: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:21:37.843697: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.849226: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:21:37.855153: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.860335: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:21:37.865737: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.870198: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.879374: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.884583: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.918780: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:21:37.929713: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:37.935029: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:38.039506: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.040663: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.041465: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:38.147978: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.148634: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.149210: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:38.171087: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 05:21:38.176139: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 05:21:38.181469: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:38.186345: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:21:38.291059: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.292042: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.292674: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:38.398690: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.399354: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.399968: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:38.510294: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.511337: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.511954: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:21:38.738115: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:21:38.739059: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:21:38.739681: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 05:21:38.802287: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_128_unit_4_Adam_with_decay_ReduceLROnPlateau/assets

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1920)
(778, 8)
 72/195 [==========>...................] - ETA: 0s

2023-09-12 05:28:10.493638: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:28:10.494706: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:28:10.495351: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 1s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[72  2  4 12  4  2  1  1]
 [ 1 76  3  7  1  4  0  0]
 [ 1  2 74  2  7  2  7  5]
 [ 6  6  1 64  5  9  4  3]
 [ 4  0  3  3 75  1  3  7]
 [ 0  2  1 15  5 65  5  3]
 [ 1  0  6 10  3  5 70  3]
 [ 6  6  5  1  7  4  2 69]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.79      0.73      0.76        98
           1       0.81      0.83      0.82        92
           2       0.76      0.74      0.75       100
           3       0.56      0.65      0.60        98
           4       0.70      0.78      0.74        96
           5       0.71      0.68      0.69        96
           6       0.76      0.71      0.74        98
           7       0.76      0.69      0.72       100

    accuracy                           0.73       778
   macro avg       0.73      0.73      0.73       778
weighted avg       0.73      0.73      0.73       778

#### Model setup:
Extractor = 'resnet_gavg'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('resnet_gavg')

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                82360     
                                                                 
 dense (Dense)               (None, 7)                 77        
                                                                 
=================================================================
Total params: 82,437
Trainable params: 82,437
Non-trainable params: 0
_________________________________________________________________

0

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()

2023-04-18 14:17:35.997639: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.
2023-04-18 14:17:38.060782: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.

Epoch 1/30

2023-04-18 14:18:04.634226: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:428] Loaded cuDNN version 8401
2023-04-18 14:18:09.438610: I tensorflow/compiler/xla/service/service.cc:173] XLA service 0x7fe738139aa0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-04-18 14:18:09.438634: I tensorflow/compiler/xla/service/service.cc:181]   StreamExecutor device (0): GeForce GTX 1080, Compute Capability 6.1
2023-04-18 14:18:10.340722: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:268] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-04-18 14:18:16.252059: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 14:18:16.830459: I tensorflow/compiler/jit/xla_compilation_cache.cc:477] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

  29/2193 [..............................] - ETA: 8s - loss: 1.9701 - accuracy: 0.1724

2023-04-18 14:18:17.270479: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 14:18:17.270549: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 14:18:17.277652: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 14:18:17.278782: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.

2193/2193 [==============================] - ETA: 0s - loss: 1.9237 - accuracy: 0.1846

2023-04-18 14:18:25.551468: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.
2023-04-18 14:18:26.056003: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 717946880 exceeds 10% of free system memory.

Epoch 1: val_loss improved from inf to 1.92166, saving model to lstm_resnet_gavg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_resnet_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('resnet_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[19  7  7 29 33  1  2]
 [ 2 72  7  9  1  1  0]
 [ 3 39 23  9 24  0  2]
 [ 5 26 16 25 20  3  3]
 [ 7 11 16 17 42  0  3]
 [ 4 40  5 34  7  6  0]
 [ 6 12 33 14 15  2 16]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.41      0.19      0.26        98
           1       0.35      0.78      0.48        92
           2       0.21      0.23      0.22       100
           3       0.18      0.26      0.21        98
           4       0.30      0.44      0.35        96
           5       0.46      0.06      0.11        96
           6       0.62      0.16      0.26        98

    accuracy                           0.30       678
   macro avg       0.36      0.30      0.27       678
weighted avg       0.36      0.30      0.27       678

#### Model setup:
Extractor = 'vgg16_avg'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_avg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 2048)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 2048)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 2048)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                82360     
                                                                 
 dense (Dense)               (None, 7)                 77        
                                                                 
=================================================================
Total params: 82,437
Trainable params: 82,437
Non-trainable params: 0
_________________________________________________________________

3948

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-04-18 14:27:48.170681: W tensorflow/tsl/framework/cpu_allocator_impl.cc:82] Allocation of 2873425920 exceeds 10% of free system memory.

Epoch 1/30
2186/2193 [============================>.] - ETA: 0s - loss: 1.4006 - accuracy: 0.5133
Epoch 1: val_loss improved from inf to 1.21263, saving model to lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 16s 6ms/step - loss: 1.3998 - accuracy: 0.5136 - val_loss: 1.2126 - val_accuracy: 0.5705 - lr: 0.0010
Epoch 2/30
2181/2193 [============================>.] - ETA: 0s - loss: 1.0341 - accuracy: 0.6522
Epoch 2: val_loss improved from 1.21263 to 1.07896, saving model to lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 12s 5ms/step - loss: 1.0337 - accuracy: 0.6523 - val_loss: 1.0790 - val_accuracy: 0.6194 - lr: 0.0010
Epoch 3/30
2185/2193 [============================>.] - ETA: 0s - loss: 0.8892 - accuracy: 0.6998
Epoch 3: val_loss improved from 1.07896 to 1.05740, saving model to lstm_vgg16_avg_10_unit_4_Adam_ReduceLROnPlateau

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_avg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 2048)
(678, 7)
170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[61  0  2 23  5  3  4]
 [ 0 64  8 11  5  3  1]
 [ 7  5 70  2 14  0  2]
 [ 8  5  8 61  3  7  6]
 [ 6  5 12 13 55  2  3]
 [ 2  6  6 31  5 44  2]
 [ 7  1  7 14  6  6 57]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.67      0.62      0.65        98
           1       0.74      0.70      0.72        92
           2       0.62      0.70      0.66       100
           3       0.39      0.62      0.48        98
           4       0.59      0.57      0.58        96
           5       0.68      0.46      0.55        96
           6       0.76      0.58      0.66        98

    accuracy                           0.61       678
   macro avg       0.64      0.61      0.61       678
weighted avg       0.64      0.61      0.61       678

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8768, 40, 512)
Shape of y_train:  (8768, 7)
Shape of X_test:  (2192, 40, 512)
Shape of y_test:  (2192, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 10)                20920     
                                                                 
 dense (Dense)               (None, 7)                 77        
                                                                 
=================================================================
Total params: 20,997
Trainable params: 20,997
Non-trainable params: 0
_________________________________________________________________

79

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30
2175/2192 [============================>.] - ETA: 0s - loss: 1.3786 - accuracy: 0.4961
Epoch 1: val_loss improved from inf to 1.10682, saving model to lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 13s 5ms/step - loss: 1.3754 - accuracy: 0.4973 - val_loss: 1.1068 - val_accuracy: 0.6191 - lr: 0.0010
Epoch 2/30
2187/2192 [============================>.] - ETA: 0s - loss: 0.9873 - accuracy: 0.6626
Epoch 2: val_loss improved from 1.10682 to 0.96288, saving model to lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 10s 5ms/step - loss: 0.9876 - accuracy: 0.6623 - val_loss: 0.9629 - val_accuracy: 0.6670 - lr: 0.0010
Epoch 3/30
2185/2192 [============================>.] - ETA: 0s - loss: 0.8779 - accuracy: 0.6965
Epoch 3: val_loss did not improve from 0.96288
2192/2192 [==============================] - 7s 3ms/step - loss: 0.8785 - accuracy: 0.6963 - val_loss: 0.9723 - val_accuracy: 0.6565 - lr: 0.0010
Epoch 4/30
2178/2192 [============================>.] - ETA: 0s - loss: 0.8067 - accuracy: 0.7221
Epoch 4: val_loss improved from 0.96288 to 0.88050, saving model to lstm_vgg16_gavg_10_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 512)
(678, 7)
170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[61  0  0 18  4  8  7]
 [ 5 71  1  7  3  4  1]
 [ 6 11 66  1 13  0  3]
 [11  9  4 60  3  6  5]
 [10  1  7 10 59  4  5]
 [ 1  2  7 18  4 62  2]
 [ 7  0  8  5 10  4 64]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.60      0.62      0.61        98
           1       0.76      0.77      0.76        92
           2       0.71      0.66      0.68       100
           3       0.50      0.61      0.55        98
           4       0.61      0.61      0.61        96
           5       0.70      0.65      0.67        96
           6       0.74      0.65      0.69        98

    accuracy                           0.65       678
   macro avg       0.66      0.65      0.66       678
weighted avg       0.66      0.65      0.66       678

#### Model setup:
Extractor = 'mobV2_gavg'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 7

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir,'mobV2_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1280)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1280)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                51640     
                                                                 
 dense (Dense)               (None, 7)                 77        
                                                                 
=================================================================
Total params: 51,717
Trainable params: 51,717
Non-trainable params: 0
_________________________________________________________________

2023-09-12 05:34:00.267972: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:00.268743: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:00.269499: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

3520

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-12 05:34:02.296789: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 1795891200 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 05:34:03.942880: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:03.943802: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:03.944440: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:04.295320: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:04.296323: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:04.297002: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:04.693247: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0xa9fa010 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-12 05:34:04.693277: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 05:34:04.693283: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 05:34:04.698245: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-12 05:34:04.806717: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2189/2193 [============================>.] - ETA: 0s - loss: 1.2585 - accuracy: 0.5994

2023-09-12 05:34:13.572357: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:13.573095: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:13.573877: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.03595, saving model to lstm_mobV2_gavg_10_unit_4_Adam_ReduceLROnPlateau

2023-09-12 05:34:14.881839: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:14.882559: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:14.883343: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:14.993999: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:14.994644: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:14.995223: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.095999: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.096851: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.097486: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.113820: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.119134: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.218144: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.218793: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.219389: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.235772: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:34:15.338710: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.339767: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.340395: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.355307: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:34:15.362687: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.368193: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:34:15.374133: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.379227: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:34:15.384637: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.389086: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.398281: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.403554: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.419221: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-12 05:34:15.429548: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.435496: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.532398: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.533061: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.533675: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.640407: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.641465: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.642091: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.673065: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:34:15.677847: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-12 05:34:15.682723: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.687286: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-12 05:34:15.787948: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.788653: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.789264: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:15.897064: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:15.898019: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:15.898641: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:16.011669: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:16.012381: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:16.013033: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:34:16.121315: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:34:16.122065: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:34:16.122809: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 05:34:16.183343: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 13s 5ms/step - loss: 1.2578 - accuracy: 0.5998 - val_loss: 1.0359 - val_accuracy: 0.6691 - lr: 0.0010
Epoch 2/30
2182/2193 [============================>.] - ETA: 0s - loss: 0.8663 - accuracy: 0.7231
Epoch 2: val_loss improved from 1.03595 to 0.90921, saving model to lstm_mobV2_gavg_10_unit_4_Adam_ReduceLROnPlateau

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1280)
(678, 7)
 56/170 [========>.....................] - ETA: 0s

2023-09-12 05:39:22.130229: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:39:22.130909: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:39:22.131554: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[70  2  0  8  9  8  1]
 [ 0 78  0  3  5  6  0]
 [ 3  4 61  3 24  2  3]
 [ 3  7  2 53 10 19  4]
 [ 0  2  5  3 72 10  4]
 [ 0  1  0 10  6 79  0]
 [ 0  0  2 14 16  4 62]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.92      0.71      0.80        98
           1       0.83      0.85      0.84        92
           2       0.87      0.61      0.72       100
           3       0.56      0.54      0.55        98
           4       0.51      0.75      0.61        96
           5       0.62      0.82      0.71        96
           6       0.84      0.63      0.72        98

    accuracy                           0.70       678
   macro avg       0.74      0.70      0.71       678
weighted avg       0.74      0.70      0.71       678

#### Model setup:
Extractor = 'dens201_gavg'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 7

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir,'dens201_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8769, 40, 1920)
Shape of y_train:  (8769, 7)
Shape of X_test:  (2191, 40, 1920)
Shape of y_test:  (2191, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                77240     
                                                                 
 dense (Dense)               (None, 7)                 77        
                                                                 
=================================================================
Total params: 77,317
Trainable params: 77,317
Non-trainable params: 0
_________________________________________________________________

2023-09-12 05:43:08.031228: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:08.032245: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:08.032867: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

0

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-12 05:43:11.043776: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.
2023-09-12 05:43:12.556095: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2693836800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-12 05:43:13.647774: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:13.648754: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:13.649453: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:14.029916: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:14.030882: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:14.031556: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:14.397917: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-12 05:43:14.474031: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7fbd200c6d50 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-12 05:43:14.474060: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 05:43:14.474067: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-12 05:43:14.479517: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-12 05:43:14.592627: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2187/2193 [============================>.] - ETA: 0s - loss: 1.2586 - accuracy: 0.5830

2023-09-12 05:43:24.018552: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:24.019547: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:24.020189: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.01012, saving model to lstm_dens201_gavg_10_unit_4_Adam_ReduceLROnPlateau

2023-09-12 05:43:25.509126: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:25.510061: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:25.510720: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:25.625419: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:25.626120: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:25.626745: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:25.733877: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:25.734756: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:25.735378: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:25.752583: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:25.757677: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:25.862208: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:25.862938: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:25.863591: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:25.881051: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:43:25.981412: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:25.982170: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:25.982854: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:25.998248: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:43:26.005940: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.011756: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:43:26.017935: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.023409: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:43:26.028986: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.033690: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.043818: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.049206: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.083417: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-12 05:43:26.094563: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.100196: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.201101: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.202007: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.202625: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:26.309073: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.309745: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.310390: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:26.341396: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 05:43:26.346324: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-12 05:43:26.351460: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.356217: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-12 05:43:26.455071: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.455943: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.456571: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:26.561263: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.561941: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.562564: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:26.672084: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.672926: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.673550: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-12 05:43:26.900749: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:43:26.901634: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:43:26.902272: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-12 05:43:26.963590: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg_10_unit_4_Adam_ReduceLROnPlateau/assets

2193/2193 [==============================] - 14s 6ms/step - loss: 1.2584 - accuracy: 0.5830 - val_loss: 1.0101 - val_accuracy: 0.6696 - lr: 0.0010
Epoch 2/30
2192/2193 [============================>.] - ETA: 0s - loss: 0.8328 - accuracy: 0.7412
Epoch 2: val_loss improved from 1.01012 to 0.95934, saving model to lstm_dens201_gavg_10_unit_4_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 1920)
(678, 7)
 73/170 [===========>..................] - ETA: 0s

2023-09-12 05:48:46.093613: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-12 05:48:46.094776: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-12 05:48:46.095571: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

170/170 [==============================] - 0s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[75  1  2  5 14  1  0]
 [ 4 77  1  2  7  1  0]
 [ 5  2 64  1 27  1  0]
 [20 12  1 31 24  8  2]
 [ 5  0  1  0 87  1  2]
 [ 2  7  0  5 19 62  1]
 [ 3  0  8  3 19  5 60]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.66      0.77      0.71        98
           1       0.78      0.84      0.81        92
           2       0.83      0.64      0.72       100
           3       0.66      0.32      0.43        98
           4       0.44      0.91      0.59        96
           5       0.78      0.65      0.71        96
           6       0.92      0.61      0.74        98

    accuracy                           0.67       678
   macro avg       0.73      0.67      0.67       678
weighted avg       0.73      0.67      0.67       678

#### Model setup:
Extractor = 'mobV2_gavg8class'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir,'mobV2_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1280)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1280)
Shape of y_test:  (2480, 8)

0

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
    gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1280)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                51640     
                                                                 
 dense (Dense)               (None, 8)                 88        
                                                                 
=================================================================
Total params: 51,728
Trainable params: 51,728
Non-trainable params: 0
_________________________________________________________________

2023-09-13 23:56:15.132561: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:15.133463: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:15.134059: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
    gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-13 23:56:20.606235: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.
2023-09-13 23:56:21.680861: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 2032844800 exceeds 10% of free system memory.

Epoch 1/30

2023-09-13 23:56:22.485216: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:22.486110: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:22.486753: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:22.832201: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:22.833212: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:22.833861: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

2472/2482 [============================>.] - ETA: 0s - loss: 1.4565 - accuracy: 0.5254

2023-09-13 23:56:33.239729: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:33.240735: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:33.241360: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.21751, saving model to lstm_mobV2_gavg8class_10_unit_4_Adam_ReduceLROnPlateau

2023-09-13 23:56:34.607924: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:34.608826: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:34.609416: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:34.721458: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:34.722141: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:34.722758: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:34.824770: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:34.825616: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:34.826238: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:34.842708: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:34.847624: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:34.948058: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:34.948751: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:34.949391: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:34.965603: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-13 23:56:35.061851: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.062695: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.063309: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.077955: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-13 23:56:35.085251: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.090692: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-13 23:56:35.096425: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.101586: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-13 23:56:35.106836: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.111284: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.120412: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.125474: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.141052: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1280]
	 [[{{node input_1}}]]
2023-09-13 23:56:35.151236: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.156512: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.253791: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.254473: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.255094: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.360195: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.361112: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.361724: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.391875: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-13 23:56:35.396544: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1280]
	 [[{{node inputs_0}}]]
2023-09-13 23:56:35.401317: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.405847: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1280]
	 [[{{node inputs}}]]
2023-09-13 23:56:35.501843: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.502525: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.503119: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.606068: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.606975: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.607606: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.715182: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.715857: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.716460: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:56:35.819521: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:56:35.820363: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:56:35.820979: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-13 23:56:35.879536: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1280]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_mobV2_gavg8class_10_unit_4_Adam_ReduceLROnPlateau/assets

2482/2482 [==============================] - 14s 5ms/step - loss: 1.4545 - accuracy: 0.5261 - val_loss: 1.2175 - val_accuracy: 0.6286 - lr: 0.0010
Epoch 2/30
2482/2482 [==============================] - ETA: 0s - loss: 1.0493 - accuracy: 0.6763
Epoch 2: val_loss improved from 1.21751 to 1.08347, saving model to lstm_mobV2_gavg8class_10_unit_4_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('mobV2_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1280)
(778, 8)
 53/195 [=======>......................] - ETA: 0s

2023-09-14 00:01:57.004383: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-14 00:01:57.005088: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-14 00:01:57.005880: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

195/195 [==============================] - 1s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[72  2  3  3  8  4  1  5]
 [ 0 75  4  2  3  4  0  4]
 [ 4  3 70  3  5  3  0 12]
 [ 7 18  4 29  7 21  3  9]
 [ 2  3 12  0 59  5  1 14]
 [ 2  6  6  1  3 73  0  5]
 [ 2  2  6  5  5 10 64  4]
 [ 2  8  4  1  4  7  1 73]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.79      0.73      0.76        98
           1       0.64      0.82      0.72        92
           2       0.64      0.70      0.67       100
           3       0.66      0.30      0.41        98
           4       0.63      0.61      0.62        96
           5       0.57      0.76      0.65        96
           6       0.91      0.65      0.76        98
           7       0.58      0.73      0.65       100

    accuracy                           0.66       778
   macro avg       0.68      0.66      0.66       778
weighted avg       0.68      0.66      0.65       778

#### Model setup:
Extractor = 'dens201_gavg8class'
Layer = '10_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'
NUM_CLASSES = 8

#del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f(data_dir,'dens201_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (9926, 40, 1920)
Shape of y_train:  (9926, 8)
Shape of X_test:  (2480, 40, 1920)
Shape of y_test:  (2480, 8)

0

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
    INPUT_SHAPE_lstm
    
    tf.keras.backend.clear_session()
    
    inputs = Input(shape=INPUT_SHAPE_lstm)
    
    # LSTM layer
    lstm = LSTM(10, return_sequences=False, dropout=0.5)(inputs)
    
    # Output layer
    outputs = Dense(NUM_CLASSES, activation='softmax')(lstm)
    
    # Create model
    model = Model(inputs, outputs)
    model.summary()
    
    mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
    reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)
    
    model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
    gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 1920)]        0         
                                                                 
 lstm (LSTM)                 (None, 10)                77240     
                                                                 
 dense (Dense)               (None, 8)                 88        
                                                                 
=================================================================
Total params: 77,328
Trainable params: 77,328
Non-trainable params: 0
_________________________________________________________________

2023-09-13 23:47:51.646801: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:47:51.647817: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:47:51.648448: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

tf.debugging.set_log_device_placement(False)
tf.config.set_soft_device_placement(True)
with tf.device('/device:GPU:1'):
    history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
    gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

2023-09-13 23:48:03.204791: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.
2023-09-13 23:48:04.796619: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 3049267200 exceeds 10% of free system memory.

Epoch 1/30

2023-09-13 23:48:05.909616: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:05.910569: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:05.911232: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:06.271508: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:06.272509: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:06.273178: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:06.665996: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600
2023-09-13 23:48:06.746199: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7fd20802f510 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-09-13 23:48:06.746227: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-13 23:48:06.746234: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (1): NVIDIA GeForce RTX 2080 SUPER, Compute Capability 7.5
2023-09-13 23:48:06.750781: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-09-13 23:48:06.859650: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

2481/2482 [============================>.] - ETA: 0s - loss: 1.5395 - accuracy: 0.4635

2023-09-13 23:48:17.145940: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:17.146964: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:17.147613: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]

Epoch 1: val_loss improved from inf to 1.20175, saving model to lstm_dens201_gavg8class_10_unit_4_Adam_ReduceLROnPlateau

2023-09-13 23:48:18.682076: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:18.683137: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:18.683753: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:18.795653: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:18.796388: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:18.797026: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:18.900731: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:18.901780: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:18.902403: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:18.919013: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:18.923932: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,?,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.025199: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.025879: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.026504: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.043080: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:48:19.139603: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.140641: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.141272: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.156196: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:48:19.163603: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.169208: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:48:19.175115: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.180431: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:48:19.185870: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.190435: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.200328: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.205589: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.239308: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'input_1' with dtype float and shape [?,40,1920]
	 [[{{node input_1}}]]
2023-09-13 23:48:19.250437: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.255961: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.356564: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.357578: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.358218: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.465899: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.466633: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.467260: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.498598: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:48:19.503487: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs_0' with dtype float and shape [?,?,1920]
	 [[{{node inputs_0}}]]
2023-09-13 23:48:19.508677: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.513449: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'inputs' with dtype float and shape [?,40,1920]
	 [[{{node inputs}}]]
2023-09-13 23:48:19.613670: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.614691: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.615305: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.722334: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.723082: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.723715: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:19.835886: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:19.836773: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:19.837408: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:48:20.057440: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:48:20.058123: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:48:20.058799: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses while saving (showing 2 of 2). These functions will not be directly callable after loading.
2023-09-13 23:48:20.117964: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'serving_default_input_1' with dtype float and shape [?,40,1920]
	 [[{{node serving_default_input_1}}]]

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_10_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_dens201_gavg8class_10_unit_4_Adam_ReduceLROnPlateau/assets

2482/2482 [==============================] - 15s 5ms/step - loss: 1.5393 - accuracy: 0.4636 - val_loss: 1.2017 - val_accuracy: 0.6028 - lr: 0.0010
Epoch 2/30
2482/2482 [==============================] - ETA: 0s - loss: 1.1148 - accuracy: 0.6371
Epoch 2: val_loss improved from 1.20175 to 1.15276, saving model to lstm_dens201_gavg8class_10_unit_4_Adam_ReduceLROnPlateau

#del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('dens201_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(778, 40, 1920)
(778, 8)
 23/195 [==>...........................] - ETA: 0s

2023-09-13 23:53:55.508082: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_2_grad/concat/split_2/split_dim' with dtype int32
	 [[{{node gradients/split_2_grad/concat/split_2/split_dim}}]]
2023-09-13 23:53:55.509184: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_grad/concat/split/split_dim' with dtype int32
	 [[{{node gradients/split_grad/concat/split/split_dim}}]]
2023-09-13 23:53:55.509824: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'gradients/split_1_grad/concat/split_1/split_dim' with dtype int32
	 [[{{node gradients/split_1_grad/concat/split_1/split_dim}}]]
2023-09-13 23:53:55.606374: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8600

195/195 [==============================] - 1s 2ms/step

(778, 8)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[72  0  1 10  4  4  3  4]
 [ 1 63  3  8  4 10  0  3]
 [ 5  1 60  2 19  3  5  5]
 [12  3  0 62  5  8  4  4]
 [ 7  0  1  4 73  3  5  3]
 [ 3  0  1 17  5 59  9  2]
 [ 6  0  3  6  4  4 73  2]
 [15  4  1  8  7  2  5 58]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.60      0.73      0.66        98
           1       0.89      0.68      0.77        92
           2       0.86      0.60      0.71       100
           3       0.53      0.63      0.58        98
           4       0.60      0.76      0.67        96
           5       0.63      0.61      0.62        96
           6       0.70      0.74      0.72        98
           7       0.72      0.58      0.64       100

    accuracy                           0.67       778
   macro avg       0.69      0.67      0.67       778
weighted avg       0.69      0.67      0.67       778

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '2_layer_128_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

del X_train, y_train, X_test, y_test
gc.collect()
X_train, y_train, X_test, y_test = read_dataset_40f('vgg16_gavg')
gc.collect()

Loaded X_train
Loaded y_train
Loaded X_test
Loaded y_test
Shape of X_train:  (8768, 40, 512)
Shape of y_train:  (8768, 7)
Shape of X_test:  (2192, 40, 512)
Shape of y_test:  (2192, 7)

0

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm1 = LSTM(128, return_sequences=True, dropout=0.5)(inputs)

lstm12 = LSTM(64, return_sequences=False, dropout=0.5)(lstm1)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm12)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 40, 128)           328192    
                                                                 
 lstm_1 (LSTM)               (None, 64)                49408     
                                                                 
 dense (Dense)               (None, 7)                 455       
                                                                 
=================================================================
Total params: 378,055
Trainable params: 378,055
Non-trainable params: 0
_________________________________________________________________

463

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30

2023-04-18 16:08:55.777980: I tensorflow/compiler/xla/service/service.cc:173] XLA service 0x7f789e69d930 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-04-18 16:08:55.803738: I tensorflow/compiler/xla/service/service.cc:181]   StreamExecutor device (0): GeForce GTX 1080, Compute Capability 6.1
2023-04-18 16:08:57.679408: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:268] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-04-18 16:09:05.347601: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:09:05.347714: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:09:05.347815: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:09:05.347946: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:09:05.705973: I tensorflow/compiler/jit/xla_compilation_cache.cc:477] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

  38/2192 [..............................] - ETA: 8s - loss: 1.9473 - accuracy: 0.2039

2023-04-18 16:09:06.291747: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.

2188/2192 [============================>.] - ETA: 0s - loss: 1.2842 - accuracy: 0.5271
Epoch 1: val_loss improved from inf to 1.14617, saving model to lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 4 of 4). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 29s 7ms/step - loss: 1.2840 - accuracy: 0.5269 - val_loss: 1.1462 - val_accuracy: 0.6099 - lr: 0.0010
Epoch 2/30
2190/2192 [============================>.] - ETA: 0s - loss: 0.9843 - accuracy: 0.6615
Epoch 2: val_loss improved from 1.14617 to 0.98768, saving model to lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 4 of 4). These functions will not be directly callable after loading.

del X_eval, y_eval
gc.collect()
X_eval, y_eval = read_evaldataset_40f('vgg16_gavg')

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

(678, 40, 512)
(678, 7)
170/170 [==============================] - 1s 2ms/step

(678, 7)

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

Confusion Matrix: 
 [[66  1  3 12  2  6  8]
 [ 2 69  4  9  3  3  2]
 [ 7  5 63  3 16  2  4]
 [10  8  4 58  9  5  4]
 [ 5  4  8  5 62  7  5]
 [ 3  4  5 11  8 62  3]
 [ 6  1  6 10  8  2 65]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.67      0.67      0.67        98
           1       0.75      0.75      0.75        92
           2       0.68      0.63      0.65       100
           3       0.54      0.59      0.56        98
           4       0.57      0.65      0.61        96
           5       0.71      0.65      0.68        96
           6       0.71      0.66      0.69        98

    accuracy                           0.66       678
   macro avg       0.66      0.66      0.66       678
weighted avg       0.66      0.66      0.66       678

#### Model setup:
Extractor = 'vgg16_gavg'
Layer = '2_layer_256_unit'
Optimizer = 'Adam'
Scheduler = 'ReduceLROnPlateau'
BatchSize = '4'

INPUT_SHAPE_lstm = (X_train.shape[1], X_train.shape[2])
INPUT_SHAPE_lstm

tf.keras.backend.clear_session()

inputs = Input(shape=INPUT_SHAPE_lstm)

# LSTM layer
lstm1 = LSTM(256, return_sequences=True, dropout=0.5)(inputs)

lstm12 = LSTM(128, return_sequences=False, dropout=0.5)(lstm1)

# Output layer
outputs = Dense(NUM_CLASSES, activation='softmax')(lstm12)

# Create model
model = Model(inputs, outputs)
model.summary()

mcp = ModelCheckpoint(f'lstm_{Extractor}_{Layer}_{BatchSize}_{Optimizer}_{Scheduler}', save_best_only=True, monitor='val_loss', mode='min', verbose=1)
reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.000001)

model.compile(loss='categorical_crossentropy',optimizer='adam' ,metrics=['accuracy'])
gc.collect()

Model: "model"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 input_1 (InputLayer)        [(None, 40, 512)]         0         
                                                                 
 lstm (LSTM)                 (None, 40, 256)           787456    
                                                                 
 lstm_1 (LSTM)               (None, 128)               197120    
                                                                 
 dense (Dense)               (None, 7)                 903       
                                                                 
=================================================================
Total params: 985,479
Trainable params: 985,479
Non-trainable params: 0
_________________________________________________________________

8390

history = model.fit(X_train, y_train, batch_size=4, epochs=30, validation_data=(X_test, y_test), callbacks=[mcp, reduce_lr])
gc.collect()

plt.plot(history.history["accuracy"])
plt.plot(history.history['val_accuracy'])
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title(f"lstm_{Extractor}_{Layer}_{Optimizer}_{Scheduler} Accuracy-Loss Chart")
plt.ylabel("Accuracy / Loss")
plt.xlabel("Epoch")
plt.xticks(np.arange(0, 30, 1))
plt.yticks(np.arange(0, 2.8, 0.2))
plt.legend(["Accuracy","Validation Accuracy","loss","Validation Loss"])
plt.show()
gc.collect()

Epoch 1/30
   1/2192 [..............................] - ETA: 1:22:28 - loss: 1.9183 - accuracy: 0.2500

2023-04-18 16:31:45.525447: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:31:45.564058: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.
2023-04-18 16:31:45.645006: W tensorflow/compiler/xla/stream_executor/gpu/asm_compiler.cc:115] *** WARNING *** You are using ptxas 10.1.243, which is older than 11.1. ptxas before 11.1 is known to miscompile XLA code, leading to incorrect results or invalid-address errors.

You may not need to update to CUDA 11.1; cherry-picking the ptxas binary is often sufficient.

2188/2192 [============================>.] - ETA: 0s - loss: 1.2926 - accuracy: 0.5291
Epoch 1: val_loss improved from inf to 1.08287, saving model to lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 4 of 4). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

2192/2192 [==============================] - 19s 8ms/step - loss: 1.2933 - accuracy: 0.5291 - val_loss: 1.0829 - val_accuracy: 0.6077 - lr: 0.0010
Epoch 2/30
2184/2192 [============================>.] - ETA: 0s - loss: 1.0040 - accuracy: 0.6545
Epoch 2: val_loss improved from 1.08287 to 0.99383, saving model to lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau

WARNING:absl:Found untraced functions such as lstm_cell_layer_call_fn, lstm_cell_layer_call_and_return_conditional_losses, lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses while saving (showing 4 of 4). These functions will not be directly callable after loading.

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

INFO:tensorflow:Assets written to: lstm_vgg16_gavg_2_layer_128_unit_4_Adam_ReduceLROnPlateau/assets

y_pred = model.predict(X_eval, batch_size=4, verbose=1)
y_pred.shape

print('Confusion Matrix: \n',confusion_matrix(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))
print('\n\nClassification Report: \n',classification_report(np.argmax(y_eval, axis=1), np.argmax(y_pred, axis=1)))

170/170 [==============================] - 1s 3ms/step
Confusion Matrix: 
 [[64  2  2 14  1  6  9]
 [ 1 69  5  6  6  5  0]
 [ 7  4 73  4  7  1  4]
 [12 10  7 52  7  6  4]
 [ 5  1 11  7 65  3  4]
 [ 6  5  6 15  6 53  5]
 [ 9  1  7  9  2  3 67]]


Classification Report: 
               precision    recall  f1-score   support

           0       0.62      0.65      0.63        98
           1       0.75      0.75      0.75        92
           2       0.66      0.73      0.69       100
           3       0.49      0.53      0.51        98
           4       0.69      0.68      0.68        96
           5       0.69      0.55      0.61        96
           6       0.72      0.68      0.70        98

    accuracy                           0.65       678
   macro avg       0.66      0.65      0.65       678
weighted avg       0.66      0.65      0.65       678

Import Libraries¶

Set Constants¶

Load Dataset¶

Load Train and validation from processed numpy¶

40fps¶

Load Evaluation from processed numpy¶

Running on Different model setup¶

40 f¶

Setup 1: Default¶

Resnet GAVG¶

Vgg16 AVG¶

Vgg16 GAVG¶

Vgg16 GAVG 8 Class¶

MobV2 GAVG¶

MobV2 GAVG 8 Class¶

MobV2 GAVG 9 Class¶

MobV2 GAVG 10 Class¶

Dens201 GAVG¶

Dens201 GAVG 8 CLASS¶

Dens201 GAVG 9 Class¶

Dens201 GAVG 10 Class¶

Setup 2: Batch size 16¶

Resnet_gavg¶

VGG16_avg¶

VGG16_gavg¶

MobV2 GAVG¶

Dens201 GAVG¶

MobV2 GAVG 8 Class¶

Dens201 GAVG 8 Class¶

Setup 3: Adam with weight decay¶

Resnet50 GAVG¶

VGG16_avg¶

VGG16_gavg¶

MobV2 GAVG¶

Dens201 GAVG¶

MobV2 GAVG 8 Class¶

Dens201 GAVG 8 Class¶

Setup 4: LSTM unit 10¶

Resnet_gavg¶

VGG16_avg¶

VGG16_gavg¶

MobV2 GAVG¶

Dens201 GAVG¶

MobV2 GAVG 8 Class¶

Dens201 GAVG 8 Class¶

Setup 5: 2 Layer LSTM¶

VGG16_gavg¶